The role of glycoconjugates as receptors for insecticidal proteins

Abstract Bacillus thuringiensis (Bt) proteins are an environmentally safe and effective alternative to chemical pesticides and have been used as biopesticides, with great commercial success, for over 50 years. Global agricultural production is predicted to require a 70% increase until 2050 to provide for an increasing population. In addition to agriculture, Bt proteins are utilized to control human vectors of disease—namely mosquitoes—which account for >700 000 deaths annually. The evolution of resistance to Bt pesticial toxins threatens the progression of sustainable agriculture. Whilst Bt protein toxins are heavily utilized, the exact mechanisms behind receptor binding and toxicity are unknown. It is critical to gain a better understanding of these mechanisms in order to engineer novel toxin variants and to predict, and prevent, future resistance evolution. This review focuses on the role of carbohydrate binding in the toxicity of the most utilized group of Bt pesticidal proteins—three domain Cry (3D-Cry) toxins.


Introduction
Bacillus thuringiensis (Bt) is a Gr am-positiv e bacterium that pr oduces a large variety of insecticidal δ-endotoxins during sporulation.These proteins may be lethal to insects and/or nematodes yet are innocuous to vertebrates and plants.Additionally, Bt proteins demonstrate species-specific activity, allowing for the eradication of harmful pests that destroy crops and spread disease without exterminating beneficial insect species.Bt proteins are an envir onmentall y safe and effecti ve alternati v e to c hemical pesticides and have now been used as biopesticides for over 50 years.In addition, genes encoding Bt proteins have been incorporated in crops such as corn and cotton with huge commercial success (Sandhu et al. 2020 ).The exact mechanisms behind Bt protein(s) toxicity are unknown, and increasing understanding is critical for the development of new Bt proteins, and to counteract emerging field resistance.
Bt pesticidal proteins may be produced during sporulation (crystal and cytolytic proteins) or the vegetative growth phase and ar e gener all y or ganized into a number of categories based on structural families, according to a recently revised nomenclature (Cric kmor e et al. 2021 ).The 3D-Cry toxins form the largest known gr oup and ar e also the most mec hanisticall y well-c har acterizedespecially those that are le pidopteran acti ve.Following ingestion by inv ertebr ates, 3D-Cry activity is proposed to occur by either of two models; the most-widely known sequential binding poreforming (Schnepf and Whiteley 1981, Bravo et al. 2007, Rodriguez-Almazan et al. 2009 ) or the alternative G-protein mediated apoptotic signalling pathway model (Zhang et al. 2006, Castella et al. 2019, Mendoza-Almanza et al. 2020 ).In the sequential binding model, Cry crystals are solubilized in the specific pH and physi-ological conditions of the insect gut, producing monomeric protoxins .T he monomers are subsequently activated by host proteinases, yielding activated Cry proteins, which bind target receptors on the brush border membranes of midgut epithelial cells.This is follo w ed b y cleav a ge within the α-helical domain I by host proteinases, triggering toxin oligomerization to form a prepore structure necessary for insertion into the phospholipid bilayer to form a channel.This culminates in cell death via colloid-osmotic lysis .T here is increasing evidence that other routes to pore formation via receptor binding may exist and that the sequential binding model may not be a universal pathway (Vachon et al. 2012, Endo et al. 2022, Sun et al. 2022 ).The signalling model differs in that there is no pore insertion, with cell death induced, instead, via the activation of an apoptotic signalling cascade-although this is not a widely accepted hypothesis.
Although significantl y differ ent at the amino acid le v el, active 3D-Cry proteins have a characteristic conserved 3-domain arc hitectur e (D-I-D-III) indicativ e of a similar mec hanism of action.Crystal structur es ar e av ailable for a n umber of acti vated 3D-Cry (Cry1Aa (Gr oc hulski et al. 1995), Cry1Ac (Derbyshir e et al. 2001 ), Cry2Aa (Morse et al. 2001 ), Cry3Aa (Heater et al. 2020 ), Cry3Bb1 (Galitsky et al. 2001 ), Cry4Aa (Boonserm et al. 2006 ), Cry4Ba (Boonserm et al. 2005 ), Cry5Ba (Hui et al. 2012 ), Cry7Ca1 (Jing et al. 2019 ), and Cry8Ea1 (Guo et al. 2009 ) along with a number of mutant and chimeric forms) and all show a conserved structur al arr angement.Domain I is linked to pore formation and consists of a helical bundle with a central hydrophobic helix-α5, associated with initializing membrane insertion, encapsulated by six amphipathic helices.Domains II and III are associated with receptor binding and ar e β-sheet-ric h domains r esembling lectins.
Both domains present structural homology to carbohydrate binding pr oteins, suc h as lectin jacalin and sialidase, r espectiv el y.This structural similarity implies that carbohydrate residues may play a critical role in receptor binding for 3D-Cry proteins-although the exact mechanisms by which this occurs remain somewhat unknown.The 3D crystal structure of the Cry1Ac1 protoxin has recentl y been elucidated, pr esenting four cysteine-ric h pr odomains (D-IV-D-VII) (Evdokimov et al. 2014 ).Domains IV and VI are alpha helical bundles that resemble spectrin or bacterial fibrinogenbinding complement inhibitor, whilst D-V and D-VII are beta-rolls that closely resemble the carbohydrate-binding moieties seen in sugar hydrolases of Family 6 carbohydrate binding module-and similar to that seen in D-II and D-III.Aside from a few recent investigations (Zghal et al. 2017, P ena-Car dena et al. 2018 ), prodomain studies hav e lar gel y indicated that it is dispensable for insecticidal activity, and instead has roles in optimizing crystal formation, pac king differ ent toxin v ariants into the same crystal, stability, selective solubilization, and ensuring sync hr onous delivery through oligomerization (Luthy and Ebersold 1981, Hofte and Whiteley 1989, Evdokimov et al. 2014 ).
Cry pr oteins ar e usuall y highl y selectiv e to their tar get insect orders, and it is unusual to find a Cry protein that effectiv el y tar gets mor e than one order-although exceptions exist, such as Cry2Aa, which has activity against Lepidoptera (Donovan et al. 1988 ) and Diptera (Yamamoto and Mclaughlin 1981b ), and Cry1Ba which has been shown to target Hemiptera (Fernandez-Luna et al. 2019 ), Lepidoptera (Simpson et al. 1997 ), and Diptera and Coleoptera (Zhong et al. 2000 ).As well as the unique domain structure in individual Cry pr oteins, tar get selectivity is determined by the presence of the receptor proteins and lipids in the target insect midgut.A relatively strong understanding of this process has been derived in Lepidoptera, where several protein types have been identified to function as Cry receptors, including; cadherin-like proteins (CAD; Nagamatsu et al. 1998, Vadlamudi et al. 1993, 1995, Gahan et al. 2001 ), GPI-anc hor ed aminopeptidases (APN; Sangadala et al. 1994, Gill et al. 1995, Rajagopal et al. 2002, Knight et al. 2004 ), GPI-anc hor ed alkaline phosphatases (ALP; Sangadala et al. 1994 , Jurat-Fuentes andAdang 2004 ), and ABC transporters (Sato et al. 2019 ).Similar receptors have been identified in other orders , e .g. mosquitoes (Diptera) utilize cadherins (Cry4Ba, Cry11Ba, and Cry11Aa), APNs (Cry11Ba), and ALPs (Cry11Aa).A series of more recent work has identified that glycosphingolipids (GSLs) can also function as Cry5B and Cry14A receptors and mediate toxicity in the nematode Caenorhabditis elegans (Griffitts et al. 2003(Griffitts et al. , 2005 ) ).
Resistance de v elopment a gainst insecticidal toxins is a common phenomenon, and a wide array of resistance mechanisms has been identified from both laboratory and field studies (Peterson et al. 2017 ).The most common mechanism appears to be altered Cry binding to receptors (Ferre and Van Rie 2002 ).Cadherins hav e r eceiv ed substantial attention due to their commonality as lepidopter an r eceptors and major mutations causing significant resistance to Cry1Ac have been identified in m ultiple str ains of Heliothis virescens (Gahan et al. 2001 ), Pectinophora gossypiella (Morin et al. 2003, Tabashnik et al. 2004, 2005, Fabrick and Tabashnik 2012, Fabrick et al. 2014 ), and Helicoverpa armigera (Xu et al. 2005, Yang et al. 2006, Zhao et al. 2010, Zhang et al. 2013 ), yet it is clear that cadherin binding and expression can be identical between resistant and susceptible strains (Siqueira et al. 2006, Bel et al. 2009 ).T his , alongside other studies, has led to the common hypothesis that a combination of other putative Cry binding moieties , such as APNs , ALPs , GSLs , and so on, ma y be r equir ed for full toxicity.
This r e vie w will focus on a ppr aisal of the liter atur e surr ounding the r ele v ance of carbohydr ate moieties in eliciting the insecticidal action of 3D-Cry proteins.In addition to the aforementioned Cry5B and Cry14A, there is ample precedent for the role of glycoconjugates as receptors for protein toxins-as is the case for cholera toxin (Holmgren et al. 1975, Kabbani et al. 2020 ), aer ol ysin (Abr ami et al. 2002 ), shiga toxin (Smith et al. 2006 ), and ricin (Sandvig et al. 1976 ).To understand how Cry toxins exploit carbohydrate moieties for toxicity in more detail, we will also provide a beginner's ov ervie w to the current understanding of the structural diversity, biosynthesis, and function of insect glycoconjugates, as well as comparing insect glycopatterning to the better-characterized pathways and glycoconjugate species present in mammals.

Gl ycoprotein gl ycans in insects and nematodes
The addition of an oligosaccharide chain to a protein backbone (glycosylation) is an extremely common posttranslational modification in eukaryotes.A substantial array of studies have concluded that gl ycopr otein moieties play critical roles in cell signalling, cell migration, cell-cell interactions, blood group determination, and immune cell trafficking-with changes in Nglycosylation associated with diverse disorders including cancers (Kodar et al. 2012 ), Crohn's disease (Verhelst et al. 2020 ), and diabetic kidney disease (Bermingham et al. 2018 ).The distinct and div er gent gl ycosylation patterns observ ed ar e driv en b y an or chestra of glycosidases and glycosyltransferases, which differ in terms of substrate specificity, and both temporal and spatial expression.The exact size and structure of the oligosaccharide can dramatically alter the biophysical properties of the pr otein-effectiv el y significantl y div ersifying the functions of a single gene pr oduct.
As with v ertebr ates, insects and nematodes demonstrate both major forms of glycosylation; N-linked (attached to Asn in an Asn-X-Ser motif, where X is not Pro) and O-linked (attached via Ser/Thr).As in mammals, insect and nematode N-linked glycosylation begins in the endoplasmic reticulum (ER) with the cotranslational transfer of a dolichol-linked precursor oligosaccharide to the aspar a gine side chain of the consensus sequence within a nascent protein.This precursor is subsequently processed in m ultiple sta ges to form matur e v ariants in the ER and Golgi.Oglycosylation also occurs in the ER, Golgi and, occasionally, the cytoplasm but unlike N-linked does not begin with a common oligosacc haride pr ecursor.
The vast majority of knowledge on insect glycoconjugates comes from the model organism Drosophila melanogaster (order Dipter a) , although ther e ar e now, collectiv el y, a gener ous number of studies on the glycomes of species within the orders Lepidoptera (Stanton et al. 2017, Cabrera et al. 2016, Fuzita et al. 2020), Hemipter a (Sc heys et al. 2019), Hymenopter a (Hyk ollari et al. 2019), and Nematoda (Cipollo et al. 2005, P asc hinger et al. 2008, Vanbeselaere et al. 2018, Wang et al. 2021 ).Genome completion of Drosophila and random mutagenesis studies have enabled the elucidation of putative genes for glycoconjugate biosynthesis and the functional impact of altering glycan patterning (Seppo andTiemeyer 2000 , Ten Hagen et al. 2009 ).

N-linked protein glycosylation
All N-gl ycans shar e the same pentasacc haride cor e, termed paucimannose (Man 3 GlcNAc 2 ),-a cor e conserv ed fr om pr otozoan to metazoan.After the dolic hol-linked pr ecursor oligosacc haride Best et al . | 3 (Glc 3 Man 9 GlcNAc 2 ) has been tr ansferr ed to the pr otein, r esident ER glucosidases and mannosidase r emov e thr ee glucose r esidues and a mannose r esidue, r espectiv el y.For most gl ycopr oteins, mannose residues are further trimmed in the Golgi generating a high mannose structure (Man 5 GlcNAc 2 ), followed by GlcNAc tr ansfer ase (GlcNAcT-1)-mediated conv ersion into a hybrid gl ycan (GlcN AcMan 5 GlcN Ac 2 ), and mannosidase II-mediated conversion into GlcN AcMan 3 GlcN Ac 2. In inv ertebr ates, this gl ycan can be trimmed further to generate paucimannose (Man 3 GlcNAc 2 ; Fig. 1 )-an N-glycan that has only rarely, and relatively recently, been detected in v ertebr ates (Lattov a et al. 2010, Balog et al. 2012, Zipser et al. 2012 ).These initial trimming stages can be follo w ed by additional enzymatic steps to add diverse sugar residues and gener ate mor e complex N-gl ycans.
Initial studies on N-linked glycans in Drosophila larvae and cultured Drosophila S2 cells sho w ed a predominance of high (Man 5 GlcNAc 2 ) and paucimannose (Man 3 GlcNAc 2 ) moieties, suggesting an absence of more complex glycans (Parker et al. 1991, Williams et al. 1991 ).These simple N-glycans can be fucosylated via α1-6 and α1-3 linkages to the reducing terminal N-GlcNac.This is div er gent fr om v ertebr ates wher e, although N-gl ycans have paucimannose as a core, the simplest N-glycan is chiefly GlcN AcMan 3 GlcN Ac 2. Furthermor e, v ertebr ates onl y fucosylate N-glycans at the α1-6 linkage.Later work, after completion of the Drosophila genome, elucidated candidate gl ycosyltr ansfer ases r equired for the generation of more complex glycans .T his , combined with impr ov ed anal ytical tec hniques, led to se v er al mass spectr ometry-based studies, whic h established the pr esence of hybrid, biantennary, and triantennary Drosophila gl ycopr oteinsincluding sulphated, glucur on ylated, and sial ylated structur es (Koles et al. 2004, North et al. 2006, Aoki et al. 2007 )-although the degree of sialyation is hotly debated (Ghosh et al. 2018, Marchal et al. 2001 ), with the only published studies re porting N-link ed sial ylated structur es at a 0.01% or unquantifiable le v el (Aoki et al. 2007, Koles et al. 2007 ).
Although simple N-glycans (Man 5 GlcNAc 2 and Man 3 GlcNAc 2 Fuc) have been predominantly observed throughout Drosophila embryogenesis, the exact profile of N-linked glycans is shown to be both spatially and tempor all y contr olled (Aoki et al. 2007(Aoki et al. , 2008 ) ).This is indicativ e of sta ge and tissue-specific gl ycopr otein r equir ements and an associated regulation of gl ycosylation mac hinery, whic h can shift the balance between paucimannose and complex structur es.Mor e than 40 distinct glycoprotein species, all containing a paucimannose core , ha ve now been identified in Drosophila, yet as observed in the earlier studies, these complex glycans are only present as minor components, with the vast majority remaining as unmodified high mannose or paucimannose structures .T his is a gain distinct fr om mammals, wher e complex N-gl ycans with abundant sial ylation ar e pr edominant.T his in v ertebr ate-specific abundance of paucimannose has been partially explained by the elucidation of a Drosophila hexosaminidase -β-N -acetylglucosaminidase, encoded by the gene fused lobes ( fdl ) (Aumiller et al. 2006, Leonard et al. 2006, Geisler et al. 2008 ).This enzyme r emov es GlcNAc residues that are added by N -acetylglucosaminyltransferase I (GlcNAcT-I), resulting in formation of paucimannose (and its fucosylated deri vati ves), whilst blocking progression to more complex glycans.Human isoenzymes ( HEXA and HEXB ) have been shown to drive paucimannosidic protein production in neutrophils (Ugonotti et al. 2022 ), through a noncanonical cascade , i. Se v er al gr oups hav e utilized mass spectr ometry to anal yse gl ycoproteins in another well-characterized model organism, the nematode C. elegans ; a body of work that has been r e vie wed in great detail by Paschinger et al . ( 2008 ).As with Drosophila , its well-c har acterized genetics helped identify candidate enzymes associated with the synthesis of hybrid and complex glycans; homologues of N -acetylglucosamin yltr ansfer ase I (Chen et al. 2002, Zhu et al. 2004 ), II (Chen et al. 2002 ), and V (Warren et al. 2002 ).Mass spectr ometric anal ysis of C. elegans N-gl ycans has shown, as in Drosophila , an abundance of high-mannosidic class gl ycopr oteins (Man 5-9 GlcNac 2 ).P aucimannosidic structur es (Man 3 GlcNAc 2 Fuc 0-3 ) are also copious in C. elegans , in which, as in Drosophila , the core can be fucosylated via α1-6 and α1-3 core linkages (Haslam et al. 2002, Paschinger et al. 2004, Cipollo et al. 2005, Natsuka et al. 2005, Hanneman et al. 2006 ).Despite the similarities, these studies also highlight se v er al distinctiv e and unique features of C. elegans N-glycan species .For example , C. elegans glycan species can be fucosylated at, up to, thr ee r esidues on the Man 2-3 GlcNAc 2 cor e and fiv e fucose r esidues on the matur e gl ycan (Fig. 1 iii) (P asc hinger et al. 2019 ).More complex C. elegans glycans can link phosphorylcholine (PC) groups to a core or terminal GlcNAc.This modification is thought to be r elativ el y fr equent in the gl ycopr oteins of C. elegans and other nematodes compared to other inv ertebr ates (Stanton et al. 2017, Martini et al. 2019 ), and associated with immunomodulatory properties (Harnett et al. 1998, Pineda et al. 2014 ) and/or be related to nematode growth and de v elopment (Loc hnit et al. 2005 ).Longitudinal studies in C. elegans have noted the N-glycan profile was distinct at each dev elopmental sta ge studied, and an incr eased degr ee of N-gl ycan complexity and PC-presence in the L1 and Dauer stages-C.elegans stages associated with significant lifestyle changes (Cipollo et al. 2005 ).Roughly 150 different N-glycan species have been identified in C. elegans and, as with Drosophila , the r elativ e pr oportion of higher order glycans is low, suggestive of a gene acting in a homologous way to the Drosophila fdl .Recent studies have shown that mutant C. elegans with a partial deletion of a β-Nacetylhexosaminidase ( hex-2 ) , pr oduce pr oportionall y less paucimannose (Gutternigg et al. 2007 ), although significant amounts are still detectable, indicating the existence of supplementary C. elegans β-N -acetylhexosaminidase genes ( hex-3 , -4 , -5 ) .
Considering the number of N-glycan structures identified, alongside the potential modifications, the structural N-glycan diversity in insects and nematodes is vast, as is the repertoire of associated roles and locations (cell surface, ion channels, adhesion, and extracellular matrix among others).In fact, a part fr om the lack of sialylation, structural diversity is reported as comparable to that of mammals (Walski et al. 2017 ).Furthermore, interspecies diversity is also clear.This is highlighted by a recent comparati ve stud y showing minimal ov erla p in the N-gl ycopr otein pr ofiles from four phylogenetically diverse insecta; the flour beetle ( Tribolium castaneum ; Coleoptera), the silkworm ( Bombyx mori ; Lepidopter a), the honeybee ( A pis mellife ra ; Hymenopter a), and the fruit fly D. melanogaster (Dipter a) (Vandenborr e et al. 2011 ).The r ele v ance of this diversity is yet to be fully understood with many questions remaining on establishing synthetic pathwa ys , determining the functional r ele v ance of N-gl ycans, and understanding the spatio tempor al contr ol thr oughout a life cycle.Indeed, shifts in glycoconjugate expression could play an important role in determining species susceptibility to a range of glycoconjugate binding toxins.In insects and nematodes, high-mannose oligosaccharides are attached in the ER to consensus Asn residues and are subsequently processed by glycosidases and glycosyltransferases to generate a variety of N-linked structures .T he synthetic pathway shown begins with the ER gl ycosidase-pr ocessed high mannose glycan (Man 5 GlcNAc 2 ).N-link ed di v ersity is limited thr ough the expression of a hexosaminidase (Fdl in Drosophila , hex-2 C. elegans ) generating paucimannose-one of the most predominant N-linked glycans in all c har acterized insecta.A common feature of insecta N-glycans-core fucosylation, occurs at C3 and/or C6 of the reducing terminal GlcNAc (via FucT6 and FucTA in Drosophila ).N-linked diversity is expanded through the expression of less well-defined glycosyltransferases (and potentially sial yltr ansfer ases).Significant div ersity and unique gl ycan signatur es hav e been noted in complex N-gl ycans between differ ent species .For example , nematode N-glycans can contain structures (i) with a bisecting galactose, (ii) with multiple phosphocholine (PC) residues as antennal modifications, and (iii) that are fucose rich with O-methylation (Me) modifications and the extension of core fucosylated residues.Nematode structures are based on figur es fr om P asc hinger et al. ( 2008), Haslam et al. ( 2002 ), and Wilson and P asc hinger ( 2016 ).Dipter an N-gl ycans with example antennae modifications (as found in Aedes aegypti , Anopheles gambiae , and D. melanogaster ; Kurz 2015 , P asc hinger and Wilson 2020 ) and an L. dispar zwitterionic lepidopteran N-gl ycan (P asc hinger and Wilson 2020, Stanton 2017 ) show featur es fr equentl y found in man y insect species: sulphated r esidues (S ±) and glucur onic acid attached to a galactose residue.Glycans are depicted according to the Symbolic Nomenclature for Glycans, as shown in KEY , linkages are shown next to the bonds, and known enzymes are named next to initial N-glycan trimming stages.

O-linked protein glycosylation
O-linked glycan diversity appears to be one of the most varied sets of posttranslational modifications across organisms and begins with the initial monosaccharide moiety linked to the (gl yco)pr otein via the oxygen atom of serine or threonine (O-S/T).These initial monosaccharides can be O-Xyl, O-Glc, O-GalNAc (mucin-type), O-Man, O-GlcNAc, or O-Fuc (Fig. 2 A-F).Mucin-type O-linked glycosylation appears to be the predominant form in Drosophila (the best-c har acterized insect species), for which the core structures and associated biosynthetic stages ar e conserv ed in v ertebr ates (as shown in Fig. 2 C).Mucintype glycans can be categorized by different core structures.In Drosophila , unmodified core-1 structures (Gal β1-3GalNAc α1-O-S/T or the 'T-antigen') are predominant (North et al. 2006 ).Core-1 structures modified with glucuronic acid (GlcA), core-2 structures (GlcNAc β1-6(Gal β1-3)GalNAc α1-O-S/T), and a less wellc har acterized HexN Ac-GalN Ac cor e structur e ar e also pr esent in a compar ativ el y r educed abundance (wher e Hex = an y six carbon monosaccharide) (Aoki et al. 2008, Breloy et al. 2008 ).Lectin binding and mass spectrometry-based characterization of the Ogl ycan pr ofiles in lepidopter an (Sf9 fr om Spodopter a frugiper da , Mb from Mamestra brassicae , and Tn from Trichoplusia ni ) and dipteran (S2 from D. melanogaster ) cell lines (Thomsen et al. 1990, Lopez et al. 1999 ), as well as larv ae fr om two mosquito species ( Aedes aegypti and Anopheles gambiae ) (Kurz et al. 2015 ) have also all demonstrated a prevalence of mucin-type core 1 and 2 structures.
As with N-glycans, an extension of the cor e O-gl ycan structur e to generate more complex patterning appears to be proportionall y r educed in arthr opods-in comparison to their mammalian counter parts (Fristr om and Fristr om 1982, Kr amer ov et al. 1996, Theopold et al. 2001, North et al. 2006 ).Further structural complexity and species-specific diversity is ac hie v ed thr ough post synthetic modifications.For example, glucur on ylated and sulphated O-gl ycans ar e observ ed in Drosophila (Br eloy et al. 2008 ), Ae. aegypti , An. gambiae , and v arious lepidopter an cell lines (Fig. 2 Gi) (Garenaux et al. 2011, Gaunitz et al. 2013 ), and phosphoethanolamine is linked to HexNAc residues in wasps and mosquitoes (Fig. 2 Giii) (Garenaux et al. 2011, Kurz et al. 2015 ).Insect and nematode gl ycan div ersity could also be heavily influenced by the environment.Indeed, cell media composition has been indiciated to influence the O-glycosylation potential of a range of insect cell lines significantly (Lopez et al. 1999 ), and an upregulation of mucins (a gl ycopr otein class wher e > 50% hav e O-gl ycosylation), has been reported in the nematode Laxus oneitus under conditions of anoxia (P ar edes et al. 2022 ).The exact role of the environment and substr ate scav enging in the r ole of insect gl ycan synthesis r emains to be determined.
Our understanding of the most common O-glycans (O-GalNAc, mucin-type) has been significantly aided through the elucidation of 14 putative Drosophila UDP-GalNAc:Polypeptide Nacetylglucosamin yltr ansfer ases (pgants)-homologs of the mammalian enzymes r equir ed for the initial tr ansfer of GalNAc fr om the UDP-GalNAc to the Ser/Thr hydroxyl group (Gerken et al. 2008, Ten Hagen et al. 2003a , b ).Bioc hemical anal ysis has shown functional conservation between mammalian and Drosophila orthologues with some pgants acting as glycopeptide transferases (GalNAc modified substrate) and others as peptide transferases (unmodified peptide substr ate).Additionall y, pgant genes ar e shown to be spatially and temporally regulated throughout Drosophila de v elopment, suggesting a distinct r egulation of Oglycan patterning (Tian and Ten Ha gen 2006 ).Demonstr ating the functional importance of a ppr opriate O-gl ycosylation, pgant35A Drosophila mutants sho w embry onic, larval, and pupal lethalitythe first demonstration of O-linked m ucin-type gl ycosylation be-ing essential for viability (Ten Hagen andTr an 2002 , Sc hwientek et al. 2002b ).Further studies with pgant35A maternal mutants sho w ed reduced localization of mucin-type glycans on the apical and luminal surfaces of the de v eloping r espir atory system and a loss of tr ac heal integrity (Tian and Ten Hagen 2007 ).Lethality is also observed in Drosophila that cannot generate the core-1 T antigen-(C1GalTa enzyme m utants)-potentiall y due to abnormalities in CNS morphogenesis (Lin et al. 2008, Xia et al. 2004 ).
Alternati ve O-link ed structures (O-Man, O-Glc , O-GlcNAc , O-Fuc , and O-Xyl; Fig. 2 ) ha ve been detected in Drosophila (Kurz et al. 2015 ), mosquitoes (Kurz et al. 2015 ), nematodes (Vanbeselaere et al. 2018 ), lepidopteran cell lines (Lopez et al. 1999 ), and hymenopter an tissues (Gar enaux et al. 2011 ), demonstr ating div ergent structures with distinct tissue distributions.Genetic studies investigating the effects of reduced transferase activity have repeatedl y demonstr ated the importance of this, more minor, glycan patterning (K ell y and Hart 1989, Ju and Cummings 2002, Okajima et al. 2003, Ten Hagen 2003a , b ) and the conservation of functional pathw ays betw een eukary otes.For example, Drosophila have two orthologues of the v ertebr ate O-mannosyltr ansfer ases (dPOMT1 and dPOMT2), encoded by rotated abdomen (rt) and twisted , (tw), whic h ar e both r equir ed for the mannosylation of pr otein substr ates (Ic himiy a et al. 2004, Ly alin et al. 2006 ).Mutations in either Drosophila rt or tw, causes defective muscular development and, as the name suggests, a rotated abdomen phenotype.In humans, mutations in Pomt genes are associated with muscular dystrophies (Muntoni et al. 2004a , b ), highlighting the functional similarities of v ertebr ate and insect O-gl ycans.As another important example, O-linked fucose (and elongated b3-linked GalNAc generated via Fringe ) residues are shown to play critical roles in embryonic de v elopment in insects and mammals through the gl ycosylation of Notc h r eceptors and subsequent modification of Notc h r eceptor ligand pr efer ences (Okajima and Irvine 2002, Okajima et al. 2003, Sasam ur a et al. 2003, Pandey et al. 2019 ).O-Xyl modification of serine residues represents the first stage in the synthesis of gl ycosaminogl ycan (GAG)-like O-glycans-linear pol ysacc harides consisting of a repeating two sugar-unit consisting of a six-carbon acidic sugar (HexA) and an amino sugar (HexNAcHexA) n .Nematodes, C. elegans and O. dentatum , have shown conservation of the common mammalian tetr asacc haride core (GlcA β1-3Gal β1-3Gal β1-4Xyl β-O-Ser) (Yamada et al. 1999, Guerardel et al. 2001 ), and also shown the addition of galactose and PC (Vanbeselaere et al. 2018 ).These nematode GAGs are demonstrated to be important for development, with the mutation of C. elegans xylosyltr ansfer ases ( sqv-2 and sqv-6 ) inhibiting GAG biosynthesis, altering vulv al mor phogenesis and zygotic cytokinesis, and maternal-effect lethality (Hwang et al. 2003 ).GAGlike gl ycans hav e also been identified in Drosophila (Yamada et al. 2002 ), and have been associated with development and facilitating pathogen invasion (Park et al. 2003, Baron et al. 2009 ).
As with N-glycans, the elucidation of currently unknown insect biosynthetic enzymes will help us to dissect the molecular function of O-glycans and the relevance of various structural features.

Glycolipids in insects and nematodes
Gl ycolipids ar e lipids with a carbohydrate attached via a glycosidic bond, with known roles in maintaining cellular membrane integrity, facilitating cell-to-cell and intracellular signalling, initiating host immune responses, and determining blood groups.GSLs are a subclass of glycolipid where the carbohydrate group is cov alentl y attac hed to a cer amide bac kbone moiety (a sphinganine , i.e .amide linked to a fatty acid; Fig. 3 ).GSLs are of particular interest when considering potential receptor functions, as they are known toxin receptors (Geny and Popoff 2006 ), and found enriched in cellular membrane microdomains (lipid rafts) that act as specialized platforms for signal transduction and protein/lipid transport (Simons andIkonen 1997 , Brown andLondon 1998 ).
Initial investigations into insect GSLs in 1973 by Luukkonen et al . ( 1973 ), sho w ed an absence of complex GSLs in cells cultured from Aedes albopictus .Ho w ever, later reports identified the first GSLs in arthropods, by utilizing 2D high-performance thin-layer c hr omatogr a phy (HPTLC) to indicate the presence of glucosylceramide (GlcCer) and mannosyl-glucosylceramide (Man-GlcCer) in two closel y r elated dipter an species; the larv ae of the gr een-bottle fly, Lucilia caesar , and the pupae of the blowfly, Calliphora vicina (Sugita et al. 1982a, Dennis et al. 1985b ).This was followed by sever al gr ound-br eaking studies fr om Sugita, Hori, Dennis, Wiegandt and others, pr edominantl y in the same dipteran species, showing arthropods form an 'arthro-series' of GSLs derived from a single, neutral, Man β1,4Glc β-ceramide core-termed mactosylceramide (MacCer) (Sugita et al. 1982a, b , 1989, 1990, Dennis et al. 1985a, b , Dabrowski et al. 1990, Weske et al. 1990, Helling et al. 1991 ).T his in v ertebr ate-specific gl ycolipid signatur e is conserv ed in nematodes and insects but is div er gent fr om v ertebr ates, wher e the majority of GSLs are derived from a lactosylceramide core (Lac-Cer; Gal β1,4Glc β-ceramide).Using a combination of HPTLC, sequential exoglycosidic digestion, methylation analysis, and directinlet mass spectrometry (MS), these aforementioned studies in dipteran insects went on to find neutral, acidic, and zwitterionic GSLs with increasing complexity and oligosaccharide length-all as extensions of the MacCer cor e. Dipter an GSLs were also identified to be fr equentl y modified with phosphoethanolamine ( P Etn) linked to C6 of GlcNAc, resulting in a zwitterionic core structure.
Drosophila melanogaster has become the pr edominant c hoice for studying arthropod GSLs, with the biosynthesis pathways and structur al v ariants now r elativ el y well-understood (Fig. 4 )-as summarized in greater detail by Aoki and Tiemeyer ( 2010 ).Analysis of Drosophila GSLs indicated the presence of a similar family of variants to that observed previously in L. caesar and C. vicina (Fredieu and Mahowald 1994, Callaerts et al. 1995, D'Amico and Jacobs 1995, Seppo et al. 2000 ).Howe v er, ther e ar e noted Drosophila distinctions such as an incr eased pr oportion of longer GSLs that are substituted with two P Etn residues (Itonori et al. 2005 , Aoki andTiemeyer 2010 ), and a 4-linked GalNAc (as opposed to a 3-linked GalNAc) in the longest c har acterized Drosophila GSL (Seppo et al. 2000 ).Studies in other insects and nematodes have also indicated that a distinct species-specific GSL diversity is present (Fig. 4 i-iv).For example, although the MacCer core is most commonly extended with GlcNAc via a β1-3 linkage follo w ed b y GalNAc via a β1-4 linkage, Drosophila can extend with Gal, rather than GalN Ac, follo w ed b y Glucuronic acid (GlcA) (Fig. 4 i) (Aoki and Tiemeyer 2010 ).Additionally, the later steps of biosynthesis a ppear to div er ge between dipter ans ( Drosophila and Calliphora ) and nematodes.In both these dipteran genera, the common cor e tetr asacc haride (GalN Ac β1-4GlcN Ac β1-3Man β1-4Glc β-Cer) is extended by a GalNAc, whereas C. elegans extends with an α1,3linked Gal.Furthermor e, the cor e GlcNAc can be substituted with PC (Fig. 4 iii)-a modification that appears to be conserved in parasitic nematodes (Gerdt et al. 1999, Wuhrer et al. 2000 ).Whether these distinctions al ways r eflect true species-specific GSLs or dev elopmentall y r egulated expr ession in the material studied (embryonic , larvae , or pupae) is not completely clear.Indeed, GSL synthesis is highly regulated in mammals-both spatially and tempor all y-with dysr egulation pr e v alent in disease suc h as stora ge disorders (Br eiden and Sandhoff 2019 ) and cancers (Furukawa et al. 2019 ).The ability to div ersify functional lipids significantl y, early in the biosynthesis pathwa y, ma y tailor GSLs for specific spatial or temporal functions-such as development or toxin binding in localized regions of the insect gut.Temporal artificial manipulation of GSL biosynthesis may be a useful a ppr oac h for investigating toxin binding at differ ent sta ges in an insect's life cycle.For example, many mammalian studies have utilized small molecule inhibitors of glycolipid biosynthesis pathwa ys , and different cell culture media ad diti ves are known to drastically alter cellular glycosylation profiles.
The presence of insect gangliosides (GSLs that contain one or more sialic acid residue) remains controversial, as reviewed pr e viousl y (Ghosh et al. 2018, Marchal et al. 2001 ).Whilst little is known about insect sial ylation, eukaryotic sial ylation is wellstudied and has diverse roles in development of the central nervous system, immune response, cell death, cell signalling pathwa ys , host-virus interaction, as well as pathogenic implications in Alzheimer's disease and cancer pr ogr ession (Varki et al. 2008, Schauer 2009, Ghosh et al. 2015, Yanagisawa et al. 2015, Teppa et al. 2016 ).Sialic acids, sialylated macromolecules and sialyltransfer ase (ST) enzymes hav e been r eported in a r ange of insects including B. mori (Kajiura et al. 2015 ), D. melanogaster (Koles et al. 2004 ), Ae. aegypti (Cime-Castillo et al. 2015, Di et al. 2017), and Galleria mellonella (Karacali et al. 1997 ) but, despite this, insect investigations indicate that gangliosides do not appear to be intrinsicall y pr esent at a detectable le v el (Aoki et al. 2007 (Wiegandt et al. 1992 ). GlcA carries a negativ e c har ge under physiological conditions, pr ompting comparisons to the sialic acid-containing gangliosides of v ertebr ates and the term 'arthr osides'.Curr entl y, ther e ar e v ery limited data to support a functional comparison.Furthermore, sialic acids can be α2-8 linked to additional sialic acids whereas GlcA dimers, to the best of our knowledge , ha ve not been reported.
In addition to the sugar component of GSLs, it must also be noted that the ceramide (a sphingoid base backbone linked to a fatty acid) backbone composition also differs between invertebrates and mammals.Mammalian sphingoid bases tend to be longer (gener all y C18) (Sullards et al. 2003 ), whereas insect sphingoid bases are generally reported as C14 and C16 and are amide linked to shorter fatty acid chains (Oswald et al. 2015 ) (Fig. 3 A and  B).In man y arthr opods, cer amide phosphoethanolamine (CPE) is the bulk sphingolipid (Fig. 3  As with v ertebr ates, the complexity of insect and nematode GSLs occurs along common biochemical pathways via specific, gl ycosyltr ansfer ase-catal yzed, sequential addition of monosaccharides.Elucidation, and manipulation, of these glycosyltransferases has provided an insight into GSL function and utility.The first committed step in GSL synthesis is through the addition of glucose to ceramide via glucosylceramide synthase (GlcCer).Knockdown of an embryonic Drosophila GlcCer homolog caused incr eased a poptosis, indicating a r equir ement for GSLs-at least during de v elopment (K ohyama-K oganeya et al. 2004 ).Catalyzing the second and third steps in Drosophila GSL synthesis are two genes brainiac (brn) and egghead (egh) -initially proposed to act in the same functional pathway based on similar de v elopmental phenotypes exhibited by their r espectiv e m utants-namel y an ov er pr olifer ation of neur al cells and enlar ged peripher al nerv es.The brn gene was determined to encode a β1,3GlcNAc tr ansfer ase directed to transfer GlcNAc preferentially to the Man β1,4Glc core structure (Muller et al. 2002, Schwientek et al. 2002a ), and egh to encode a β1,4-mannosyltr ansfer ase to form MacCer (Fig. 4 ) (Wandall et al. 2003 ).Both Brn and egh mutants are lethal, implying a requirement for second and third step sugar addition.Inter estingl y, inhibiting the fourth step in GSL synthesisvia null mutation of β1,4 -N -acetlygalactosaminyltransferases ( β4GalNAcTB/ β4GalNAcTA) is not lethal, although still causes defects including the v entr alization of ov arian follicle cells (Chen et al. 2007 ).Drosophila α1,4-N -acetylgalactosaminyl transferase Figure 4. Ov ervie w of GSL synthesis and diversity in insects and nematodes.Arthroseries GSLs present in insects and nematodes are built around a common MacCer cor e, gener ated thr ough addition of Man to a glucosylcer amide, in contr ast to the mammalian GlcCer cor e, exampled in GM3 (i, red box).This core can be extended to form more complex structures, such as those shown abo ve , which ha ve all been identified in Drosophila embryos (and in some cases, other Diptera).Drosophila figure components modified from Sharrow et al. ( 2010 ).Modification with phosphoethanolamine ( P Etn) on GlcNAc residues is frequent and generates zwitterionic GSLs, whilst addition of GlcA to terminal Gal residues generates acidic groups.Some of the known gl ycosyltr ansfer ases that facilitate GSL biogenesis ar e marked; Drosophila ( Egh , Brn , α4GT1, β4GalNAcTA, and β4GalNAcTA) and C. elegans (BRE-5, BRE-3, and BRE-4).Although the structur es in the 'Arthr o-series' bo x with the sequential ad dition of GlcN Ac then GalN Ac appear to be the most common root of more complex GSLs across invertebrate species, species-specific variants are frequently noted, such as those depicted in (ii) B. mori (Itonori 2018 ), (iii) C. elegans (Griffitts 2005 ), (iv) Drosophila (Aoki and Tiemeyer 2010 ), and (v) C. vicina (Dennis et al. 1985a ).PC = phosphorylcholine, gl ycans ar e depicted according to the Symbolic Nomenclatur e for Gl ycans as shown in KEY .
( α4GTI) synthesizes the ceramide-pentahexoside (Mucha et al. 2004 ), although as fourth step ( β4GalNAcTB/ β4GalNAcTA) mutants are still viable, this is also presumably nonessential for viability.Toxicity studies in the nematode C. elegans (discussed in greater detail below) have found genes homologous to brainiac and egghead , bre-5 , and bre-3 , r espectiv el y.
As with N and O glycans, it is clear that an increasing range of gl ycolipid structur al v ariants is being identified in insects and nematodes, e v en if these more complex structures do not make up the majority of the total pool.Key to deciphering the molecular function of these glycoconjugates is the elucidation of glycosyl-tr ansfer ases.Altering gl ycolipid biosynthesis pathways-thr ough manipulation of gl ycosyltr ansfer ase activity via gene silencing or inhibitory compounds-will help to inform a ppr oac hes to w ar ds curr ent, and nov el, methods of pest control.

Glycoconjugates as membrane receptors for insecticidal and nematocidal toxins
The role of host cell membrane glycoconjugates as toxin receptors has ample precedent (Zuverink and Barbieri et al. 2018 ).Toxins that r el y on gl ycopr otein binding include pertussis toxin (Stein et al. 1994 ) and aer ol ysin (Diep et al. 1998 ).Examples of protein toxins shown to use lipid-moieties to facilitate entry include the poreforming toxins lysenin (via sphingomyelin (Yamaji et al. 1998 ) and c holester ol-dependent cytol ysins (Tweten et al. 2005 ), Shiga toxin (via GSL Gb3; Okuda et al. 2006, Shin et al. 2009 ), and c holer a toxin (via GM1a ganglioside; Wernick et al. 2010 ).Lipid microdomains are also implicated in toxin binding due to the high concentration of GSLs present.For example, cholera toxin-induced membr ane curv atur e is shown to be dependent on both the multiplicity and specific geometry of GM1a binding sites (Kabbani et al. 2020 ), and Shiga toxin is localized to Gb3 in lipid rafts (Smith et al. 2006 ).Some toxins, such as members of the Botulinum toxin family, utilize both a ganglioside and a protein receptor, whereas others, such as ricin, bind a specific carbohydrate moiety that can be present on either a glycolipid or a glycoprotein (Fu et al. 1996 , Zuverink andBarbieri 2018 ).Below we will discuss the existing r esearc h surr ounding the r ole of gl ycoconjugates in insecticidal and nematocidal 3D-Cry protein to xin acti vity.Lectins are carbohydr ate-binding pr oteins whic h ar e, individuall y, highl y specific to a distinct sugar group (Cummings and Etzler 2009 ).Lectins have been incredibly useful, and widely used, in elucidating the sugar binding properties of various insecticidal toxins; those discussed in this r e vie w ar e summarized in Table 1 .
Se v er al of the studies, discussed below, utilize cellular models to investigate 3D-Cry binding affinity and toxicity.In these studies, it is worthwhile to consider the impact of pH, as 3D-Cry proteins are solubilized and activated in the midgut lumen due to selective pH conditions (Knowles et al. 1994 ).In the liter atur e, the insect midgut is often r eferr ed to as alkaline-a c har acteristic , i.e .often cited to assist in conferring insect species selectivity.Indeed, the majority of Dipteran and Lepidopteran species assessed have an alkaline midgut ( ∼pH 8.0-10.0),although there are exceptions such as Marasmia trapezialis (pH 7.0-7.2),Pieris rapae (pH 7.3-7.6),and Corcyra cephalonica (pH 7.0-7.6)(Berebaum et al. 1980 ). Furthermor e, ther e ar e often differ ences between the posterior and anterior midgut r egions, suc h as Ae.aeg ypti and Aedes canadensis mosquito larvae ( ∼pH 8 in the gastric caecum, > pH10 in the anterior midgut, pH 7.5 in the posterior midgut) (Dadd et al. 1975, Boudko et al. 2001 ).In contrast, other insects can have a mildly acidic midgut such as Coleoptera, Leptinotarsa decemlineata (pH 6.5-5.36)(Krishnan et al. 2007 ) and Diabrotica virgifera virgifera (pH 5.75) (Kaiser-Alexnat 2009 ).In terms of cell culture experiments the pH will be determined by buffer or culture media (whic h ar e fr equentl y mor e acidic than mammalian media,∼pH 6.2-6.5).In many experiments the toxin in question is solubilized and activ ated befor e addition to cells, via extr acted 'midgut-juice' or artificially with buffer and proteinases-which in theory should negate the need for 'mid-gut' conditions for solubilization and activation but may alter the binding affinities via protonation states of k e y residues.

Binding to BBMVs show Cry1Ac binds in a GalNAc-dependent manner
The Cry1A subclass of le pidopteran-specific to xins are of great commercial importance and the most well-studied 3D-Cry toxins .T he earliest glycoconjugate binding studies were performed using endotoxin isolated from Bt serovar.kurstaki HD-1 (Btk HD-1), which was later confirmed to contain three distinct Cry1A proteins that share > 76% aa identity as protoxins; Cry1Aa, Cry1Ab, and Cry1Ac (Hofte and Whiteley et al. 1989 ).These early stud-ies proposed the occurrence of a common Cry insecticidal poreforming action (Hofmann et al. 1988a , b ), yet identified mechanistic heterogeneity dependent on individual Cry pr oteins, tar get species, and putative binding 'receptors'.Of note, early studies using the Btk HD-1 strain also likely contain other Cry proteins including Cry2Aa2, Cry2Ab2, and Cry1Ia3.
The r ele v ance of gl ycoconjugates in eliciting to xin acti vity was r ecognized earl y on, with Kno wles et al .( 1984 ) sho wing that GalNAc and GlcNAc binding-lectins (SBA and WGA, r espectiv el y) neutralized activity of lepidopter an-activ e δ-endotoxin proteins fr om Btk str ain HD-1 in a lepidopter an cell line (CF1) isolated from the Cry1A-susceptable cabbage butterfly ( Choristoneura fumiferana ).Using the same model, they went on to identify the first putative Cry 'receptor'-a 146-kDa cell-surface glycoprotein capable of binding both SBA and δ-endotoxin (Knowles and Ellar 1986 ). Dennis et al. ( 1986 ) first proposed that glycolipids were responsible for modulating δ-endotoxin actions, through demonstrating Btk HD-1 toxin binding to distinct C. vicina pupal GSLs-of which some species contained a r ele v ant terminal GalNAc residue.In these studies, they isolated both total neutral and total acidic gl ycolipid fr actions, and isolated neutr al GSL components that they probed using a thin layer chromotography (TLC) o verla y technique to detect binding of both the protoxin and activated forms of Btk HD-1 proteins.Although Btk HD-1 contains a number of toxins (Yamamoto and McLaughlin 1981a ), the authors only used the ∼130 kDa proteins-most lik ely re presenting a mix of Cry1 proteins.Multiple binding partners were observed in both glycolipid fractions, with the main component (bound by both the protoxin and activated forms) being Gal α1-3GalN Ac β1-4GlcN Ac β1-3Man β1-4Glc β1-4Cer (denoted as 5B by the authors).Although both the protoxin and activated form were shown to bind strongly to the Gal-terminal 5B glycolipid, the toxin sho w ed a decrease in binding specificity after activation, with an increased number of glycolipids bound and an increase to w ar ds glycolipids with terminal GalNAc r esidues.Differ ent binding patterns between the pro and active forms would indicate the binding of protoxin would not block activity of the activated protein through competition for binding.When reading these works, it is important to consider that these binding experiments utilized models containing cells deriv ed fr om nontar get tissues, whic h potentiall y pr esent gl ycoconjugates found pr edominantl y outside of the midgut, and in an altered abundance.Brush border membrane vesicles (BBMVs) pr epar ed fr om larv al midguts pr ovided a mor e ' in vivo' r epr esentation and became common in the field for investigating toxin binding to apical microvilli.Using BBMVs or gut tissues, isolated fr om a r ange of lepidopter an species, se v er al inv estigations confirmed a range of specific Cry1A binding sites with nM affinity constants (Jaquet et al. 1987, Van Rie et al. 1989, 1990, Wolfersberger et al. 1990, Ferre et al. 1991, Garczynski et al. 1991, Denolf et al. 1993 ).In many cases the le v el of Cry1A toxicity w as sho wn to correlate with binding affinity (Hofmann et al. 1988b ;Van Rie et al. 1989, 1990, Garczynski et al. 1991, Denolf et al. 1993 ).For example, Cry1Ab and Cry1Ac recognize the same receptor on Ostrinia nubilalis BBMV, yet the former has an 11-fold higher affinity whic h corr elates with a 10-fold higher toxicity (Denolf et al. 1993 ).The importance of these binding sites was further illustrated by work in a field population of Plutella xylostella , where resistance to Cry1Ab was associated with loss of BBMV binding sites (Ferre et al. 1991 ).Furthermore, these studies illustrated frequent receptor heterogeneity and the existence of multiple binding sites, with increased binding site concentration also associated with increased toxicity (Van Rie et al. 1989, 1990, Garczynski et al. 1991 ).For example, H. virescens larvae show three different populations of binding site, one which binds Cry1Aa, Cry1Ab, and Cry1Ac, a second which binds Cry1Ab and Cry1Ac, and a third restricted to Cry1Ac binding.This correlates with the pronounced larvicidal differ ence between Cry1A v ariants (Ac > Ab > Aa) (Van Rie et al. 1989, 1990 ). Receptor pr oteins originall y identified fr om ligand binding studies in BBMV have since been purified and c har acterized.Two major forms of putative Cry receptor have been identified, namely cadherin-like receptors (CAD) (Vadlamudi et al. 1993(Vadlamudi et al. , 1995 ) ), and aminopeptidase-N (APN) famil y r eceptors (Knight et al. 1994, Sangadala et al. 1994 et al. 2019 ).Roles for putativ e gl ycosylation sites in the latter two r eceptor families ar e less well-explor ed-with no specific role for gl ycosylation r eported for Cry1 ABC r eceptors.
Ho w e v er, toxicity does not al ways corr elate with BBMV protein binding (Van Rie et al. 1990, Wolfersberger et al. 1990, Ferre et al. 1991, Garczynski et al. 1991 ).This is exemplified by Garczynski et al . ( 1991 ), showing similar high affinity Cry1A binding to BBMVs isolated from both highly susceptible ( Manduca sexta and H. virescens ), moder atel y susceptible ( Helicoverpa zea ), and tolerant ( S. frugiperda ) lepidopteran larvae .Kumaras wami et al. ( 2001 ), andHiguchi et al. ( 2007 ), demonstrated BBMV proteins isolated from either susceptible or resistant populations of P. xylostella have the same Cry1A binding ca pacity, yet r esistant insectderived BBMV and gut tissue had a significant reduction in neutral GSLs, indicating these glycolipids can mediate toxin susceptibility .In resistant P .xylostella populations, this was accompanied by decreased oligosaccharide length, with synthesis arrest at the pentasacc haride sta ge and a slightl y r educed activity of Gal and GalNAc tr ansfer ase, suggesting that mor e elabor ate gl ycolipid moieties facilitate Cry1A toxicity (Kumaraswami et al. 2001 ).Mor e r ecent w ork b y Ma et al. ( 2012a ), supports the r ole of gl ycolipids in Cry1Ac binding and tolerance.Helicoverpa armigera larvae demonstr ate enhanced toler ance to Cry1Ac if they ar e pr efed with LEC-8-a galectin-like protein isolated from nematodes.Both LEC-8 and Cry1Ac w ere sho wn to bind to gut glycolipids in a similar manner, implying that LEC-8 inhibits Cry1Ac glycolipid binding sites, thus mediating tolerance.The LEC-8 natural ligand is unknown, but an inhibitory ELISA showed lactose can inhibit LEC-8 binding to H. armigera gut glycolipids by 20%, and a mild inhibitory effect was observed with GalNAc , galactose , mannopyranose, inositol, and trehalose.LEC-8 has also been shown to interact with Asialofetuin-a glycoprotein with terminal GalNAc residues (Nemoto-Sasaki et al. 2008 ).
Differ ences in neutr al sugar content between susceptible and resistant M. sexta populations has been reported to correlate with Cry1A binding by a number of groups (Sangadala et al. 2001, Jurat-Fuentes et al. 2002 ).Knowles et al. ( 1991 ), solidified a role for a glycoconjugate in Cry1A binding in insect gut epithelia.GalNAc addition completely abolished Cry1Ac binding in M. sexta , partially in H. virescens , but had no effect on Pieris brassicae .This correlated with SBA and Cry1Ac binding the same (gl yco)pr otein in M. sexta and H. virescens , but not P. brassicae , collectiv el y indicating GalNAc is a component of the Cry1Ac receptor(s) in some lepidopteran species, but gl ycopr otein inter action is not r equir ed in others e.g.P. brassicae.The authors did not investigate the possibility of binding to GalNAc present in glycolipids.Although much of the literature to date is focused on the role of GalNAc in eliciting Cry1Ac toxicity, Haider and Ellar ( 1987 ) have proposed the relevance of D-Glc in eliciting Cry1 activity.Here, the authors sho w ed the activity of a trypsinized lepidopteran-specific preparation from Bt ser ov ar.aizaw ai IC1 (containing a 55-and a 58-kDa polypeptide) is completely inhibited in M. brassicae cells by D-Glc and the D-Glc binding lectin-ConA.It is not clear exactly what protein toxins wer e expr essed in this pr epar ation, although Cry1Ab7 is r eported in this strain (Haider and Ellar 1988 ).
Conv ersel y, gl ycolipid and sugar binding is also implicated in enhancing tolerance to Cry toxins through the sequestration of toxin oligomers in the gut and subsequent pr e v ention of r eceptor binding in the midgut brush border (Ha yaka wa et al. 2004, Ma et al. 2012a , b ).The peritr ophic membr ane (PM) is the semipermeable lining of the insect midgut which, among its functions, acts as protection fr om mec hanical and pathogenic dama ge.Se v er al studies have indicated that compromising the integrity of the PM can enhance Bt toxin activity in insect larv ae, pr esumabl y thr ough allowing more insecticidal protein to reach receptors at the midgut epithelium brush border (Granados et al. 2001 ).Ha yaka wa et al .( 2004 ), demonstrated that the interaction of Cry1Ac with the PM can be inhibited with the addition of GalNAc in the Cry1Ac tolerant lepidopteran species , B .mori .Upon addition of GalNAc , Cry1Ac passes through the PM significantly quicker, and at a similar rate to the B. mori acti ve to xin, Cry1Aa-although the authors did not demonstrate if this renders B. mori Cry1Ac susceptible.Ma et al. ( 2012b ), have suggested that binding of Cry toxin to glycolipids in lipophorin (lipoprotein particles that transport lipids in insect haemol ymph) incr eases Cry toxin tolerance .T hey demonstrated that D-II of Cry1Ac monomers binds glycolipids from lipophorin particles, and forms Cry1Ac oligomers in the presence of glycolipids isolated from both H. armigera and G. mellonella cell-free plasma and midgut tissue.Cry1Ac addition to G. mellonella lipid particles induced a ggr egation-an inter action thr ough whic h, the authors suggest, Cry1Ac is sequestered to the gut lumen.This study also used TLC to show the main Cry1Ac glycolipid binding species present in H. armigera gut tissue migrated to a similar position as globoside Gb4 (GalNAc 3 β1-2Gal α1-4Gal β1-4Glc β1-1-Cer)-which has a terminal GalNAc.
The exact mechanistic basis for Cry1A toxicity remains unclear.A large body of data shows insecticidal activity is dependent on m uc h mor e than a single r eceptor inter action, but with the exact insect system, toxin oligomerization state, multicomponent complexes, and tissue localization all having profound effects on toxicity.The most established mechanism for Cry1A appears to be that of sequential binding during which a toxin monomer is recognized by a cadherin-lik e rece ptor causing a conformational c hange, whic h facilitates pr epor e oligomer formation (and distinct types of pr epor e may be possible e v en for the same toxin; Gomez et al. 2014 ), and the subsequent binding to APN enabling membrane insertion.Multiple and complex receptor binding is not uncommon in the toxin field outside of 3D-Cry proteins, e.g.diphtheria (Hasuwa et al. 2001 ) and pr otectiv e antigen (Scobie et al. 2003 ) are determined to utilize more than one receptor.Furthermore, as discussed in the introduction, the role of the prodomains in toxicity is yet to be fully elucidated.Aside from the commonly hypothesized roles in toxin stability, formation, and stabilization (Derbyshire et al. 2001 ), the structure of Cry1Ac1 protoxin D-V and D-VII have four predicted ligand binding sites for galactose , N -acetylglucosamine , mannose , and xylose (Zghal et al. 2017 ), presenting the possibility that D-V and D-VII could inter act with gl ycans in the gut, and ma y be in volv ed in pr otoxin recognition of a receptor.In support of this idea, a recent study b y P eña-Car deña et al. ( 2018), has demonstrated the C-terminal protoxin domain of Cry1Ab provides additional binding sites for ALP and APN r eceptors, r esulting in a higher binding affinity of the pr otoxin, whic h corr elates with incr eased toxicity-compar ed to the activated form.

APN and APN gl ycosyla tion in mediating Cry1A binding and activity
Utilizing protoxin affinity chromatography and anion-exchange c hr omatogr a phy, (Knight et al . 1994 ), purified a gl ycopr otein (APN1) present in the midgut target tissue of M. sexta that was bound by Cry1Ac and SBA, but not Cry1B.Sequencing of the bound gl ycopr otein r e v ealed sequence similarity to the APN familya heavil y gl ycosylated zinc aminopeptidase , i.e .a common feature of the insect midgut and, ther efor e, often used to assess BBMV purity.APNs have since been extensively studied as Cry receptors and many different lepidopteran variants have been c har acterized-although not all bind Cry pr oteins.APNs ar e divided into eight phylogenetic classes (Cr av a et al. 2010, Hughes et al. 2014, Fonseca et al. 2015 ), with single insect species able to expr ess m ultiple r eceptors fr om differ ent classes.APN isoforms that bind Cry1Aa (Masson et al. 1995 ), Cry1Ab (Masson et al. 1995, Denolf et al. 1997), and Cry1Ac (Gill et al. 1995, Valaitis et al. 1995, Wang et al. 2005b, Luo et al. 1997, Nakanishi et al. 2002, Angelucci et al. 2008 ) have been discov er ed in multiple lepidopteran species-although curr ent e vidence suggests onl y Cry1Ac binds via GalN Ac, recognized b y moieties present in a surface cavity in D-III, i.e. not conserved in Cry1Aa or Cry1Ab (Burton et al. 1999, de Maagd et al. 1999b, Jenkins et al. 2000, Masson et al. 1995 ).Putative Cry toxin receptors have been identified in APN classes 1-5, although r ecentl y APNs fr om classes 6 and 8 hav e been implicated in mediating toxicity of Cry1Ab, Cry1Ac, and Cry1Ca in Chilo suppressalis larvae (Sun et al. 2020 ).
The crystal structures of Cry1Ac and Cry1Ac in complex with GalNAc have been published (Fig. 5 ) (Derbyshire et al. 2001 ) and although this has provided evidence of D-III involvement in GalNAc binding, exactl y wher e the GalNAc receptor ligand is located on APN is unknown.Sequence analysis of class 1 Cry1Ac-binding M. sexta APN isoforms sho w ed the presence of 4-7 potential N-linked glycosylation consensus sites and 13 putative O-glycosylation sites (Knight et al. 1995, 2004, Stephens et al. 2004 ).A total of 10 of the putative O-linked sites ar e pr edicted in a Thr/Pro rich region of the C-terminus, thought to form a 'stalk' that raises the active site above the membrane.Lectin recognition of these M. sexta Apn1-linked glycans indicated the presence of fucosylated and high mannose N-glycans (ConA, AAA, GNA, and UEA1 lectin binding), and O-linked glycans (SBA lectin binding) (Denolf et al. 1997, Knight et al. 2004 ).As presented in Fig. 6 , the presence of N-and O-linked glycosylation sites can be predicted by sequence analysis.Comparing the sequences of Cry-binding lepidopteran midgut APNs we see the number of N-glycosylation sites does not v ary dr amaticall y between classes (0-6 sites per protein), and the positioning of these sites is some what similar-especiall y between members of the same class .T he number of O-linked sites does differ dr amaticall y between sequences (1-46 sites), with classes 1 and 3 sequences containing substantially more consensus sites (13-46) than classes 2, 4, and 5 (1-6).Pr e vious anal ysis of lepidopteran APN sequences using an earlier version of Ogl ycosylation site pr ediction softwar e (NetOGl yc v3.1, opposed to v4.0) predicted no consensus sites for class 2 receptors (Pigott and Ellar 2007 ).
Individual species of N-linked glycoconjugates on the 120-kDa M. sexta Apn1 have been identified through MALDI-T OF/T OF tandem mass spectrophotometry coupled with lectin binding and exoglycosidase digestion.These included the common insect paucimannose structure (Man 3 GlcNAc 2 ) linked to Asn609, and highly fucosylated structures at the other three consensus sites (Asn295, Asn623, and Asn752).These glycans w ere sho wn to display up to a trifucosylated core and fucosylated antennae structures (Fuc 1-3 GlcNAc).This predominance of Fuc α1,3GalNAc-Asn is further indicated by the resistance of APN to PNGase F-an enzyme that cleaves all aspar a gine-linked oligosacc harides unless the core contains an α1,3 fucose (Stephens et al. 2004 ).It is unlikely that these high-fucose glycans are responsible for Cry1Ac binding as they lack terminal GalNAc residues, suggesting it is the C-terminal O-site glycans that might determine Cry1Ac binding.Supporting this hypothesis, Cry1Ac is not reported to bind to any class 2 lepidopteran APNs-a class which has significantl y fe wer pr edicted O-linked gl ycosylation sites and no C-terminal stalk region (Fig. 6 ) (Pigott and Ellar 2007 ).Although O-glycosylation sites have been hypothesized to be critical for Cry1Ac activity, there is evidence of Cry1Ac binding and activity in APN classes with compar ativ el y low numbers of O-glycosylation consensus sites.Cry1Ac can bind to a class 4, 110 kDa APN present in H. virescens BBMV, that does not contain a C-terminal stalk and is not recognized by SBA (Banks et al. 2001 ).Furthermore, a class 5 APN isolated from Athetis lepigone (AlAPN5) has recently been identified as a putative functional receptor mediating Cry1Ac toxicity (Wang et al. 2017b ).T his ma y indicate that the incr eased O-gl ycosylation sites seen in classes 1 and 3 are not responsible for Cry1Ac toxicity, although, to the best of our knowledge, it is unknown whether Cry1Ac binding to AlAPN5 is GalNAc-dependent.Further investigations are required to determine if glycosylation is required for Cry1Ac binding to classes 2 and 5 APNs, or if these receptors work via a GalNAc-independent route.An array of studies has shown lepidopteran APNs, of all classes, ar e attac hed to the membr ane via gl ycosyl-phosphatidylinositol (GPI) anchors (Gill et al. 1995, Knight et al. 1995, Valaitis et al. 1995, Denolf et al. 1997, Hua et al. 1998 ).GPI-anchors contain carbohydr ate-ric h structur es, often including cor e-linked GalNAc present at the membrane surface, leading to speculation that this may be a Cry1A binding epitope.Ho w e v er, r emov al of the GPIanc hor gl ycan moiety using phospholipase C (PLC) does not appear to alter binding activity (Masson et al. 1995 ), although it does dr asticall y r educe Cry1Ac por e-forming activity-as expected by loss of membrane association (Lorence et al. 1997 ).GPI-anchored pr oteins, including APN, ar e pr efer entiall y cluster ed in gl ycolipidenric hed micr odomains-specialized deter gent-r esistant membr ane micr odomains pr esent in both mammals and insects that ar e enric hed in c holester ol and GSLs.Chemical anal ysis of the 115-kDa M. sexta APN-associated lipid a ggr egate sho w ed a predominance of neutral lipids, mainly diacylglycerol and free fatty acids (Sangadala et al. 2001 ).The presence of neutral lipids is inter esting giv en the afor ementioned studies indicating a reduction in neutral GSLs in resistant populations of P. xylostella and M. sexta (Kumaraswami et al. 2001, Higuchi et al. 2007 ).Reconstitution of the 115-kDa M. sexta APN into liposomes sho w ed increased Cry1Ac binding when the lipid a ggr egate was present, as well as pr efer ential binding of Cry1Ac to lipid microdomains (Sangadala et al. 2001 ).This concentration of APNs to lipid microdomains is hypothesized to facilitate toxin oligomerization through the high density of binding epitopes.Oligomerization of Cry1Ac and Cry1Ab is shown to facilitate membrane insertion and pore formation via significantly increasing the binding affinity to APN ( ∼100-fold over the monomeric form) (Pardo- Lopez et al. 2006 ).Ne v ertheless, these lipid domains could also be r equir ed for protection from gut proteases or APN structural stabilization.Furthermor e, lipid r afts a ppear to be r equir ed for the por e-forming actions of GalNAc-insensitive Cry1Ab (Zhuang et al. 2002 ), indicating they are not simply just enhancing toxicity via increasing GalNAc receptor concentration.
The exact role of APN and glycoconjugates in facilitating Cry1A toxicity is yet to be fully understood, with several studies indicating APN binding alone is not always enough to induce toxicity.For example, Banks et al. sho w ed Drosophila S2 cells transfected with a novel 110 kDa APN from H. virescens conferred binding but did not induce pore formation (Banks et al. 2003 ).Furthermor e, r emoving APN binding does not necessarily eliminate all binding, with Lee et al . ( 1996 ), showing APN competes for Cry1Ac binding with Lymantria dispar BBMV-but does not eliminate it.Howe v er, a significant number of reports indicate APN is critical for pore-formation (Sangadala et al. 1994, Schwartz et al. 1997, Gill and Ellar 2002 ).For example, expression of the 120-kDa M. sexta APN in the mesodermal and midgut tissue of Drosophila is capable of r endering normall y insensitiv e larv ae susceptible to Cry1Ac (Gill and Ellar 2002 ).Furthermor e, se v er al studies show that artificial APN suppression confers Cry1A resistance in several Lepidoptera (Qiu et al. 2017b, Sun et al. 2020 )  the same gut, by exploring the difference in Cry1Ac binding to BBMV isolated from either the anterior (A-BBMV) or posterior (P-BBMV) midgut of a target insect, M. sexta .Cry1Ac binding to P-BBMV induced a faster rate of toxicity, compared to A-BBMV, but was substantiall y r educed by the pr esence of GalNAc, whilst A-BBMV binding was not.Furthermore, Cry1Ac binding to APN was concentrated in P-BBMV suggesting the GalNAc-sensitive mechanism involves APN, whilst the GalNAc-insensitive binding does not.Indeed, later studies by Banks et al . ( 2001 ) supported this idea showing that Cry1Ac recognized a distinct 110 kDa APN in H. virescens , where binding was not inhibited by GalNAc and the receptor itself did not bind SBA.Furthermore, a mutant Cry1Ac with an altered GalNAc binding pocket demonstrated enhanced binding to the 110-kDa APN v ariant, e v en though binding was abolished to the GalNAc-mediated 120 and 170 kDa H. virescens APN variant.
As briefly discussed abo ve , a model of how APN confers Cry1Ac toxicity is through a bivalent sequential binding mechanism, with an initial low-affinity, r a pidl y r e v ersed inter action (GalNAcsensitive) follo w ed b y a slo w er high-affinity irr e v ersible inter action (GalNAc insensitive) (Cooper et al. 1998, Jenkins et al. 2000 ).Combined mutational, binding and toxicity studies have enabled the identification of residues important for Cry1Ac binding to APN and GalNAc (Fig. 5 and Table 2 ).Br oadl y, D-I is associated with insertion of the pore into the membrane, and APN binding epitopes ar e primaril y localized to Cry1A D-II and D-III (Rajamohan et al. 1996a, b , c , Vachon et al. 2004, Liu and Dean 2006 ).Domain II has been shown to influence membrane insertion, via a high affinity interaction with APN, whereas D-III is hypothesized to be involved in host specificity and the initial low-affinity receptor recognition (Wu andDean 1996 , de Maagd et al. 1999a , b )-such as the GalNAc-dependent binding mechanism of Cry1Ac (Burton et al. 1999, de Maagd et al. 1999b, Jenkins et al. 2000 ).Indeed, sequence analysis has shown D-III to be markedly div er gent in Cry1Ac compared to other related-non-GalNAc binding-3D-Cry pr oteins (Br avo et al. 1997, Thompson et al. 1995 ).
The first phase of APN recognition is hypothesized to be through fast, low affinity D-III binding.This is supported by Lee et al . ( 1999 ), who generated a series of alanine substitution mutations in the region of D-III unique to Cry1Ac (503-525 aa) and demonstrated that binding affinity was significantl y r educed, and to a r elativ el y gr eater degr ee than toxicity (Fig. 5 B).Whilst some of these mutant residues are in direct contact with GalNAc (Q509, R511, and Y513,), others are not (S503, S504, N506, N510, and W545)-but with the exception of W545 are in close proximity to the binding pocket.It was not investigated whether any of these mutations affect GalNAc binding, making it difficult to interpret whether reduced mutant binding to BBMVs was through a loss of GalNAc binding.The authors conclude that if D-III is predominantl y involv ed in initial low-affinity APN binding, then this will onl y compr omise second phase high-affinity binding when it is r educed by at least 5-fold.Burton et al . ( 1999 ) also reported substitution mutations in the unique region of Cry1Ac D-III (N506D, Q509E, and Y513A-the latter two having direct contact with GalNAc in the crystal structure of the complex) resulted in reduced binding and slo w er pore formation, with the triple mutation no longer inhibitable by GalNAc-yet no significant differences in toxicity wer e observ ed.Further supporting that D-III binding is r equir ed for sequential D-II binding, the mutation of a tryptophan residue (W545A) in D-III (Fig. 5 B) can completely abolish sequential binding of D-II to the L. dispar APN and recognition of GalNAc-of particular note given W545 is not part of the GalNAc binding pocket (Jenkins et al. 2000 ).Inter estingl y, all Cry1Ac tryptophan residues ar e conserv ed in the closel y r elated Cry1Ab, except the D-III W545 residue (Rausell et al. 2004 ).The complete loss of APN binding in L. dispar , via the Cry1Ac W545A m utation, onl y caused a 50-fold decr ease in activity, wher eas the same W545A m utation in M. sexta larvae did not abolish binding to APN, with little to no loss in toxicity (Pardo- Lopez et al. 2006 ).The work in M. sexta also demonstrated that GalNAc binding to the Cry1Ac oligomer increases the exposure of W545 to solvent, through a subtle conformational change in the GalNAc binding pocket region of D-III.In M. sexta , this conformational change is hypothesized to be responsible for the marked increase in binding affinity of the Cry1Ac oligomer to APN.Collectiv el y, these data indicate that D-III functions to bind both GalNAc and APN in a low affinity manner, which can affect second-phase APN binding, yet there are apparent species-specific differ ences whic h determine Cry1Ac inter action with APN and toxicity, and an indication that Cry1Ac can retain toxicity e v en when binding to APN and GalNAc is abolished-leaving the binding open to further investigation.Domains II and III are not specifically linked to glycan interactions, yet a common theme is a ppar ent between mutational studies in all three domains; the binding to APN and subsequent toxicity are not necessarily correlated.This could be explained by the presence of alternative in vivo Cry1Ac r eceptors-suc h as cadherin-lik e rece ptors or ABC transporters-that function independently of APN and could be potentially compensating for the lack of APN binding/activity.The exact model used may change the distribution/concentration of APN and any potential alternativ e r eceptors.Furthermor e, the exact experimental setup may play a significant role.If APN binding to D-III is the rate limiting step to binding to D-II, and D-II binding and membrane permeabilization is not abolished but slo w ed, it may be possible to exert toxicity over a longer time course.A better understanding of the k e y r esidues in Cry1Ac r equir ed for binding to receptors, and the role of GalNAc in this binding, might enable improved engineering of both insect specificity and toxicity, as well as providing a valuable tool for identifying potential resistance-driving mutations.Vadlamudi et al . ( 1995 ), purified and characterized the first cadherin-lik e rece ptor fr om M. sexta larv ae, a 210-kDa pr otein termed BT-R 1 .Sequence analysis sho w ed a 30%-60% similarity to the cadherin superfamily of pr oteins-a lar ge famil y of tr ansmembr ane gl ycopr oteins c har acterized by r epeated calciumbinding domains.Since the discovery of BT-R 1 , receptors with a highly similar domain organization have been identified in an array of other lepidopteran species including B. mori (BtR175), H. virescens (HevCaLP), O. nubilalis , L. dispar , P. xylostella (PxCad), C. suppressalis (CsCad), and H. armigera (HaCad) (Nagamatsu et al. 1999, Gahan et al. 2001, Morin et al. 2003, Flannagan et al. 2005, Wang et al. 2005a, Xu et al. 2005 ).Lepidopteran cadherin receptors are usually anchored to the apical membrane of the midgut epithelium via a single tr ansmembr ane domain and, unlike GPIanc hor ed r eceptors (suc h as APNs or ALPs), ar e not pr efer entially localized to glycolipid-enriched lipid microdomains (Zhuang et al. 2002, Midboe et al. 2003 ).Inter estingl y, Cry1Ab tr eatment of M. sexta micr ovilli membr anes was shown to induce Bt-R 1 localization to lipid microdomains-although this is likely due to Bt-R 1 r emaining attac hed after toxin oligomerization and not due to a r equir ement for (glyco)lipid-facilitated binding (Bravo et al. 2004 ).

Cry1A binding to cadherin-like receptors
Ther e ar e significant data to show cadherin-like r eceptors function in determining Cry1A specificity and toxicity in lepidopteran larvae (Pigott and Ellar 2007 ) and lepidopteran and Drosophiladerived cell lines (Keeton and Bulla 1997, Hua et al. 2004, Zhang et al. 2005 ).Furthermor e, expr ession of BT-R 1 and BtR175 in mammalian-derived cell lines can induce Cry1Ac toxicity (Dorsch et al. 2002, Tsuda et al. 2003 ), suggesting cadherin-like receptors alone may be enough to permit cytocidal action and no other 'insect-specific' featur es ar e r equir ed for action.The success of inducing Cry1A toxicity in cell lines through cadherin-like receptor expression alone may be due to the redundancy of glycosylation in specifying binding.Unlike APN, there are no reports of sugars acting as binding competitors with Cry1Ac to cadherin-like receptors.Further indication that glycosylation is not r equir ed comes from a study showing that the shortest fr a gment of Bt-R 1 that binds Cry1A toxins is a nonglycosylated 169 aa ectodomain fragment, i.e. also capable of inhibiting toxicity (Dorsch et al. 2002 ).To the best of our knowledge, the current literature does not report glycosylation to play a significant role in cadherin-like receptor binding, although N-and O-linked glycosylation sites are present on all identified le pidopteran cadherin-lik e rece ptors (Shao et al. 2018 ).

Cry1A binding to ALP receptors
Selection of a Cry1Ac resistant strain of H. virescens allo w ed for comparison of midgut epithelium proteins between susceptible (YHD2) and resistant (YHD2-B) larvae (Jurat-Fuentes et al. 2002 , Jurat-Fuentes andAdang 2004 ).After observing reduced Cry1Ac binding to YHD2-B BBMVs, based on the rationale that GalNAc forms part of the Cry1Ac receptor, the authors investigated le v els of SBA binding to BBMVs and indeed observ ed r educed SBA binding to YHD2-B resistant larvae-initially indicative of altered glycosylation (Jurat-Fuentes et al. 2002 ).Further c har acterization of YHD2-B BBMVs identified a 68-kDa gl ycopr otein as a GPI-anc hor ed alkaline phosphatase-HvALP.Digestion of BBMV proteins with PNG-F to release N-terminal oligosaccharides, eliminated SBA binding to HvALP, confirming the presence of N-linked oligosaccharides with terminal GalNAc residues.Addition of Cry1Ac abolished SBA binding to HvALP, indicating competitive binding of both proteins for the same N-linked GalNAc r esidues on HvALP.Corr elating with r educed Cry1Ac binding, Cry1Ac-resistant BBMVs also demonstrated a reduction in expression and a 3-fold decrease in activity of HvALP-suggesting the resistance was not due to altered glycosylation or recognition of GalNAc, but instead due to a reduction in HvALP protein expression-although the authors did not perform oligosaccharide anal ysis, r esistance thr ough alter ed gl ycosyl inter actions cannot be completely ruled out (Jurat-Fuentes and Adang 2004 ).In a parallel with the work described abo ve , Ning et al. 2010 described two ALPs cloned from H. armigera (HaALPs) that specifically bind Cry1Ac via N-linked GalN Ac.Whether GalN Ac binding on ALP is r equir ed for Cry1Ac toxicity is still open for debateindeed GalNAc addition to H. armigera BBMVs inhibits permiabilization (Rodrigo-Simon et al. 2008 ), ho w e v er, whether this is dir ectl y thr ough ALP and the r ele v ance to in vivo activity is yet to be determined.

Cry5B and Cry14A
Cry5B is the best-c har acterized of the Cry5 subfamil y of six phylogeneticall y r elated pr oteins (Cry5Aa, Cry5B, Cry12A, Cry13A, Cry14A, and Cry21A) that may demonstrate nematocidal and/or insecticidal activity (Wei et al. 2003 ).Consistent with the mode of 3D-Cry protein insecticidal toxin actions, susceptible nematodes fed with nematocidal Bt strains experience dose-dependent lethality associated with reduced feeding activity, inhibited de v elopment and intestinal damage.To date, both Cry5B and Cry14A nematocidal activity is shown to be dependent, at least in part, on gl ycolipids (Marr oquin et al. 2000, Griffitts et al. 2001, 2003, 2005 ).
Using forw ar d genetics in C. elegans , Marroquin et al .( 2000 ) identified five bre genes (for Bacillus -toxin resistant), four of which confer high le v els of r esistance to Cry5B induced toxicity and one ( bre-1 ) that confers a significantly lo w er level.In all resistant mutants, Cry5B toxin remained in the intestine and was not internalized into the gut cells indicating resistance via reduced 'receptor' binding.The first bre gene to be c har acterized was Bre-5 , found to encode a β1,3-galactosyltr ansfer ase with str ong sequence similarity to the Drosophila brn gene (r equir ed for glycolipid synthesis; see Fig. 4 ) (Griffitts et al. 2001 ).Successiv el y, bre-2 , bre-3 , and bre-4 were characterized as encoding further glycolipid synthetic proteins; bre-4 as a UDP-GalNAc:GlcNac β1-4-N -acetl ygalactosamin yltr ansfer ase, bre-2 encodes a β1,3 glycosyltr ansfer ase, and bre-3 a putative glycosyltransferase homologous to Drosophila egh (see Fig. 4 ) (Griffitts et al. 2003(Griffitts et al. , 2005 ) ). Functional homology of bre genes to the egh-brn inv ertebr ate-specific lipid glycosylation pathway was shown via TLC lipid anal ysis, demonstr ating that bre mutants express no ( bre-3 , bre-4 , and bre-5 ), or significantl y r educed ( bre-2 ) complex GSLs, yet have no change in N-or O-linked pr oteogl ycan pr ofiles.Specific binding of Cry5B to these bre -dependent complex GSLs alongside genetic epistasis-based experiments supported the proposal that bre -genes act consecutiv el y ( bre-3 , bre-4 , bre-5 , and bre-2 ) to synthesize a functional lipidlinked oligosacc haride r eceptor with terminal galactose residues (Griffitts et al. 2005 ).In further support of GSLs as principal determinants for Cry toxicity, the C. elegans LEC-8 galectin (a ßgalactoside-binding protein) can compete with Cry5B for carbohydrate binding.Cry5B binding to C. elegans glycolipid-coated TLC plates was inhibited through the addition of recombinant LEC-8, and C. elegans LEC-8 deficient mutants were more susceptible to Cry5B, in comparison to wild type worms (Ideo et al. 2009 ).Bre m utants also demonstr ated a moder ate r esistance to Cry14A, a toxin with 34% sequence identity to Cry5B in their protoxin forms and ∼30% identity in the activated form.This r elativ el y low le v el of amino acid identity suggests that other distantly related toxins may induce bre -mediated toxicity.Ho w e v er, the r educed r esistance, compared to Cry5B, signifies that other Cry14A receptor(s) may compensate for the loss of the bre -mediated glycolipid (Griffitts et al. 2001(Griffitts et al. , 2003 ) ).
Although identified in the same forw ar d genetics screen as bre 2-5 , bre-1 mutants demonstrate substantially less Cry5B resistance Marroquin et al. 2000, Barrows et al. 2007 ).Bre-1 has since been identified as a GDP-mannose 4,6 dehydratase (GMD), an enzyme involved in a fucose salvage pathway.Unlike the bre2-5 genes, it does not function in a glycolipid-specific manner, with bre-1 defective mutants showing strikingly reduced levels of fucosylated N and O-linked pr oteogl ycans as well as fucosylated glycolipids (Barrows et al. 2007 ).This partial Cry5B resistance indicates that fucose is less critical for eliciting Cry5B binding than terminal galactose residues-as sho wn b y competitive binding studies.
Inter estingl y, no ob vious c hange in phenotype or lethality wer e observed in the bre -mutant C. elegans , apart from a small reduction in brood size in bre-1 and bre-3 worms (Barrows et al. 2007 ).The nematode is a ppar entl y ca pable of surviving with reduced levels of GSLs and dr amaticall y r educed fucose, whic h is perha ps surprising given the commonality of fucose in nematode glycans and the pr e v alence of detrimental phenotypes in mammalian GMD knoc k outs (K eeley et al. 2019(K eeley et al. , Sturla et al. 2001 ).This has implications for Cry resistance in nematodes, since they can tolerate changes in glycosylation while in Drosophila, the equivalent brn and egh mutants are lethal/sterile, suggesting a significantly lo w er toler ance to r educed bre -mediated gl ycosylation and an essential role for GSLs in insects .T his contrast in phenotypes could suggest that insects, in contrast to nematodes, would be less able to ac hie v e to Cry resistance via GSL alteration.
Sequence analysis of HaABCC1 sho w ed the presence of 14 potential N-glycosylation sites and 16 potential O-glycosylation sites thr oughout the entir e pr otein (Chen et al. 2018 ).In HaABCA2, sequence analysis identified six potential N-glycosylation sites within the extracellular (EC) domain loops of tr ansmembr ane domain (TMD)-1 and TMD-2 (Tay et al. 2015 ).One of these putative N-glycosylation sites is located within a 5-bp deletion mutation shown to confer Cry2Ab resistance in H. armigera .This deletion m utation intr oduces a stop codon within HaABCA2 TMD-2, leading to a protein truncation.Although the use of these these putativ e gl ycosylation sites is yet to be confirmed, it has been hypothesized that binding of Cry2A toxins to the glycosylated EC domain loops of ABCA2 may form the basis of toxin oligomerization and sequential pore formation (Tay et al. 2015 ).Giv en that ABC tr ansporters have been shown to exist as multiprotein complexes in the membrane , it ma y also be the case that other ABC-associated proteins are involved in Cry2A binding and pore-formation (Kaminski et al. 2006 ).Other reported Cry2 receptors include the Se-V -A TPase subunit B from S. exigua , also predicted to contain several putative glycosylation sites (Qiu et al. 2017a ).
Cry2Ab has also been shown to interact with lipophorin glycolipids (Ma et al. 2012b ).As discussed pr e viousl y, Ma et al . ( 2012b ) demonstrated that Cry1Ac addition to G. mellonella lipid particles induced a ggr egation, and sequesters Cry1Ac to the gut lumen, possibl y incr easing Cry toxin tolerance.Cry2Ab was also shown to a ggr egate following lipid particle interaction and, hence, the authors suggest a similar mechanism of toxin tolerance.

Cry3
The Cry3 class is the best-c har acterized of the coleopteranspecific proteins, with a domain architecture consistent with other 3D-Cry proteins .T he lectin-like D-III of Cry3Aa was found to exhibit strong resemblance to the N-terminal cellulose binding domain (CBD N1 ) of the bacterial Cellulomonas fimi 1,4-β-glucanase C (CenC) (Johnson et al. 1996, Burton et al. 1999 ).The CBD N1 domain of CenC has been shown to interact with cellulose, as well as cell oligosaccharides and β-1,4-linked oligomers of glucose (Tomme et al. 1996 )-with binding thought to occur via β-strands within a fiv e-str anded cleft which constitutes the CBD N1 (Johnson et al. 1996, Kormos et al. 2000 ).The structural correlation between Cry3 D-III and the CBD N1 of CenC may suggest a role for sugar moieties in Cry3 receptor binding.
Se v er al studies hav e implicated C ADs , ALPs , APNs , and ABCs as Cry3 binding proteins and/or functional receptors-although less is known regarding the r ele v ance of gl ycosylation.In Tenebrio molitor , Cry3Aa has been shown to bind to a GPI-anc hor ed ALP, whic h is pr efer entiall y expr essed in the BBMV of early instar larv ae (Zuniga-Nav arr ete et al. 2013 ).In D. virgifera virgifera and Chrysomela tremula , ABCB1 has been identified as a functional receptor for Cry3A (Niu et al. 2020 ).Functional validation of the D. vir gifera vir gifera Cry3A receptor (DvABCB1) was achieved through acti vated Cry3A ad dition to Sf9 or HEK293 cells, both expressing DvABCB1.As the gut of D. virgifera vigifera is mildly acidic, this again indicates the pH of cell studies does not need to replicate the gut environment for toxicity to occur, in the presence of acti vated to xin.Sequence analysis of CtABCB1 predicts two putativ e gl ycosylation sites on the EC loops of the tr ansmembr ane domains (P auc het et al. 2016 ).Although the functional r ele v ance of these sites is unknown, this is the first study to suggest that glycosylation may be important for ABC receptors.

Cry30Ca2
Cry30Ca2 is produced by the Bt ser ov ar.jegathesan , a mosquitocidal subspecies that shows toxicity a gainst Ae. aeg ypti , An. stephensi , Culex pipiens , and C. quinquefasciatus (Delecluse et al. 1995, Kawalek et al. 1995 ).Bioassays of the isolated Cry30Ca2 toxin indicate that this individual protein is not toxic against C. quinquefasciatus and, hence, additional studies are required to test its toxicity to other mosquitoes (Sun et al. 2013 ).
Using homology modelling, based upon Cry4Ba, Zhao et al .( 2012 ) produced a 3D model of the Cry30Ca2 structure consistent with that of other 3D-Cry proteins.Dissimilar from the interaction of Cry1Ac with GalNAc, which occurs via Cry1Ac D-III, docking studies investigating the interaction of Cry30Ca2 with GalNAc highlighted a distinct, putative binding site within the apical loops of the Cry30Ca2 lectin-like D-II (residues I321 in loop 1, Q342, T343, Q345 in loop 2, and Y393 in loop 3, which form se v en hydr ogen bonds with GalNAc) (Zhao et al. 2012 ).Various studies have implicated the loop regions of Cry protein D-II in receptor binding, including Cry3Aa, which is shown to bind TmCad1 via D-II loop 1 (Zuniga-Nav arr ete et al. 2015 ).Giv en the r esults of molecular docking studies and these structural similarities, a role for GalNAc containing carbohydrate moieties in Cry30Ca2 mosquitocidal activity has been suggested (Zhao et al. 2012 ).Ho w e v er, the effect of GalNAc on the activity of Cry30Ca2 is yet to be investigated.

Cry11a
Bt ser ov ar.israelensis (Bti) str ains ar e highl y to xic to a n umber of mosquito species and, as suc h, ar e used for the control of their populations in the field (Mittal et al. 2003 ).One such Bti toxin is Cry11Aa, which displays toxicity against Aedes and Culex larvae and, to a lesser extent, Anopheles larvae (Otieno-Ay ay o et al. 2008 ).Cry11Aa has been identified to bind receptors in mosquito larvae that are in the same classes as Cry toxins that act against Lepidoptera, including APNs, ALPs, and Cadherins.
Cry11Aa binding to an ALP is suggested to play a role in mediating toxicity in Aedes larvae (Fernandez et al. 2006 ).Inter estingl y, the interaction between Cry11Aa and Ae.aegypti ALP1 (AaeALP1) w as sho wn to be modulated b y other pr oteins-namel y C-type lectins and galectins-which both interfere with toxicity (Batool et al. 2018, Zhang et al. 2018 ).C-type lectins are a superfamily of proteins that have mannose and galactose type carbohydr ate binding ca pabilities thr ough conserv ed r esidues (Br own et al. 2018 ).Galectins are a family of proteins that typically bind to β-galactoside carbohydrates (Modenutti et al. 2019 ), although compar ativ el y little is known about their carbohydrate binding properties and function in invertebrates compared to vertebrates (Yang et al. 2011, Zhang et al. 2018 ).The Ae. aegypti C-type lectin-20 (CTL-20) can bind to both Aedes BBMVs and recombinant AaeALP1, in addition to binding to Cry11Aa itself.Further to this, CTL-20 has been shown to compete with Cry11Aa for binding to AaeALP1 suggesting that they bind AaeALP1 in the same region (Batool et al. 2018 ).Similarly, galectin-14 has been shown to compete with Cry11Aa for binding to AaeALP1 and Aedes BBMVs, with modelled molecular docking indicating that Cry11Aa and Galectin-14 bind to ALP1 on two different, but overlapping, interfaces (Zhang et al. 2018 ).Additionally, other galectins such as galectin-6 have also been shown to interfere with Cry11Aa toxicity (Hu et al. 2020 ).There is some evidence that galectin-6 binds to molecules containing galactose-β1,4-fucose (Takeuchi et al. 2008, Maduzia et al. 2011 ), ther efor e, it is possible that Cry11Aa may be able to bind similar glycan moieties .T hese results draw comparison with the Cry5B data discussed pr e viousl y, wher e LEC-8 competes for carbohydrate binding and suggests a role for glycan moieties in the interactions between Cry11Aa and ALP1.Howe v er, to the best of our knowledge, there are no studies which have directly investigated the involvement of glycan residues in this binding.
Perhaps the most extensive work looking into the role of glycoconjugates in Cry11Aa receptor binding has come from Chen et al. ( 2009a ), investigating the interactions between Cry11Aa and AaeAPN1.This study identified AaeAPN1 as a Cry11Aa binding partner through pulldown assays utilizing biotinylated toxin performed on solubilized Ae. aegypti BBMV.The AaeAPN1 was cloned and expressed in both E. coli and Sf21 cells-significant in the context of investigating the role of glycoconjugates as E. coli do not natur all y N-and O-gl ycosylate pr oteins (Du et al. 2019 ).The glycosylation status of AaeAPN1 from BBMV was investigated through lectin blots (Chen et al. 2009a ) and demonstrated the native form of AaeAPN1 was detectable by WGA but not SBA, indicating AaeAPN1 contains N -acetylglucosamine moieties but not terminal N -acetylgalactosamine residues.In Sf21 cells, expression of a catal yticall y activ e form of AaeAPN1 did not render cells susceptible to Cry11Aa treatment.Although Sf21 cells have the ability to N-and O-gl ycosylate pr oteins (Davis and Wood 1995 ), the AaeAPN1 in these cells was not detected by WGA, SBA, ligand blot, or toxin pull down assays and the band detected by anti-APN1 antibody was smaller than expected-possibly due to differences in post translational modifications (Chen et al. 2009a ).The authors also hypothesized that alternative glycosylation in Sf21 cells could mask a glycan-independent binding site.Taken together these results may indicate that that glycosyl moieties are r equir ed for binding.Ho w e v er, Chen et al. ( 2009b ) also demonstrated, via dot blot and competitive ELISA, that a truncated AaeAPN1 fr a gment expr essed in E. coli binds to Cry11Aa, suggesting that this interaction is glycan-independent, due to the absence of N and O-glycosylation in E. coli-although this does not determine whether this binding is involved in mediating toxicity.
The sequence of an Aedes cadherin protein shown to bind to Cry11Aa has been determined and there are predicted Nglycosylation sites within the cadherin repeats, ho w ever, there has been no investigation so far into whether glycosylation is present and if it is r equir ed for this interaction (Chen et al. 2009b ).

Cry4Ba
Cry4Ba is also produced by Bti and is processed in the insect midgut to produce an acti ve to xin of 65 kDa (Angsuthanasombat et al. 1991 ).Like Cry11Aa, Cry4Ba also targets Aedes and Anopheles mosquito larvae (Otieno-Ay ay o et al. 2008, Ben-Dov et al. 2014 ,), and is shown to target the same receptor classes as other 3D-Cry toxins (APNs , ALPs , and Cadherins) (Likitvivatana vong et al. 2011, Saengwiman et al. 2011 ).
A cadherin Cry4Ba binding partner (AgCad1), expressed in An. gambiae BBMVs was predicted to be glycosylated, based upon the observ ed AgCad1 pr otein band having a slightl y lar ger molecular weight than expected (Hua et al. 2008 ).The same group also demonstrated that Cry4Ba displays limited binding on dot blots to an E. coli -expressed truncated peptide from AgCad1 (a CR11 membr ane pr oximal EC domain peptide), suggesting that some binding is possible in the absence of glycosylation or other in vivo r equir ements.Similarly, Cry4Ba was shown to bind to a segment of the An.gambiae cadherin BT-R 3 , expressed in E. coli , which consisted of the EC domain module 7 through to the membrane proximal EC domain (Ibrahim et al. 2013 ).As this cadherin fr a gment was expressed in E. coli it is unlikely to be glycosylated and provides further evidence that glycosylation of cadherins is not required for Cry4Ba binding.
Multiple studies have implicated ALPs as binding partners for Cry4Ba (Bayyared d y et al. 2009, Dechklar et al. 2011, Jimenez et al. 2012 ).Muta genesis studies demonstr ated Cry4Ba binding to ALP1, in part, through D-II loop II.Multiple Cry4B D-II mutants displayed reduced binding to ALP1 from BBMV and E. coli , and reduced toxicity to Ae. aegypti larvae .T he results of this study suggest it is unlikely that receptor glycosylation is essential for interaction, as the mutated versions of Cry4Ba also display reduced binding to E. coli expressed ALP1 and Ae.aegypti BBMV (Jimenez et al. 2012 ).Further to this , T hammasittirong et al. ( 2011) sho w ed that Cry4Ba binds to an Ae.aeg ypti ALP expr essed in E. coli with high affinity, which they conclude supports the notion that Cry4Ba interactions with ALPs does not require glycosyl moieties as proteins expressed in E. coli are unlikely to be gl ycosylated.Finall y, Buzdin et al. 2002 sho w ed through ligand blots that addition of monosaccharides (mannose , glucose , galactose , galactosamine , N -acetylglucosamine , and N -acetylgalactosamine , either individually or in mixtures) did not interfere with Cry4Ba binding to ALP that was pr epar ed fr om Ae. aeg ypti BBMVs, with similar r esults shown for Cry11Aa binding to ALP.They also demonstrated that the addition of N -acetylglucosamine or N -acetylgalactosamine failed to elute ALP from Cry4Ba-and Cry11Aa-Sepharose (Buzdin et al. 2002 ).APNs have also been identified as receptors for Cry4Ba (Saengwiman et al. 2011 ).Sf9 cells expressing two Ae.ae-gypti APN isoforms (AaeAPN2778 and AaeAPN2783) displayed increased sensitivity to Cry4Ba and the toxin w as sho wn to bind to APNs in Sf9 cells (Aroonkesorn et al. 2015 ).The APNs expressed in these cells were thought not to be glycosylated, suggesting that the interaction between APNs and Cry4Ba is glycan independent.
Cry4Ba has been shown to interact directly with lipid bilayers, whic h is perha ps not sur prising giv en the elucidation of GPIanc hor ed APN and ALP receptors .T hammasittirong et al ( 2019 ) tested full length Cry4Ba and D-III-only binding to lipid bilayers and liposomes pr epar ed fr om an artificial lipid mix containing phosphatid ylethanolamine, phosphatid ylcholine and cholesterol (but no glycans).They focused on D-III of Cry4Ba as it is shown to bind along the a pical micr ovilli of the larval midgut of Ae. aegypti (Chayaratanasin et al. 2007 ).Domain III of Cry4Ba displayed tight binding to immobilized liposome membranes with a K D comparable to that of the full-length protein.Ho w ever, unlike the fulllength protein, the truncated D-III Cry4Ba fragment did not induce ion-channel formation in planar lipid bilayers or permeability of calcein dye-loaded liposomes, consistent with the role of this domain as a membrane anchor rather than having a role in pore formation (Thammasittirong et al. 2019 ).The binding of Cry4Ba to lipids may suggest that, like other Cry proteins, it localizes to lipid rafts-although whether glycolipid binding occurs, remains to be tested.

Glycan binding in other bacterially produced insecticidal toxins
There is also evidence that glycan binding could play an important role in the insecticidal action of other structural classes of bacteriall y pr oduced toxins, with lectin-like domains present in the Tpp family (D-I) (Colletier et al. 2016 ), Vegetative insecticidal pr otein famil y (Vip3, D-IV, and D-V) (Zheng et al. 2020 ), mosquitocidal holotoxin (Mtx1Aa1) (Treiber et al. 2008 ), and the membrane attac k complex/perforin famil y (Mpf, C terminal domain) (Zaitse v a et al. 2019 ).Sugar binding appears to play a role with several members of the Tpp family, including Tpp78, Tpp80, and the Tpp1/Tpp2 binary complex.Se v er al sugars-including c hitotriose, N -acetylm ur amic acid, c hitobiose, and N -acetylneur aminic acidcan reduce the mosquitocidal action of Lysinibacillus sphaericusproduced Tpp1/Tpp2 in Culex cell lines (Broadwell and Baumann et al. 1987 ), and arabinose and fucose can reduce Tpp1 toxicity to w ar ds Culex larvae (Sharma et al. 2018 ).Both galactose and GalNAc have recently been demonstrated to inhibit the activity of Bt-produced Tpp78 (Cao et al. 2022 ) and Tpp80 (Best et al. 2022 ) a gainst their r espectiv e tar gets, rice planthoppers ( Laodelphax striatellus and Nilaparvata lugens ) and mosquitoes ( C. quinquefasciatus , Ae. aegypti , and An.gambiae ).The mosquitocidal Mtx1Aa1 contains 12-putative sugar binding domains across four ricin B-type lectin r epeats, whic h ar e structur all y r elated to Piersin-a cytotoxin, i.e. reported to bind Gb3 and Gb4 glycolipids (Matsushima-Hibiya et al. 2003 ).This is just a snapshot of the glycan-binding literature on other bacterial pesticidal proteins, and highlights glycan binding as an important mechanistic theme acr oss bacteriall y pr oduced pesticidal proteins.

Conclusions
Bt 3D-Cry toxins are critical for progressing a sustainable appr oac h to controlling pests of agriculture and vectors of human disease, with the de v elopment of field r esistance thr eatening curr ent effectiv eness and pr ogr ess.Understanding the mec hanism of action is k e y to understanding resistance and the potential dev elopment of ne w 3D-Cry pr oteins.All known 3D-Cry pr oteins contain lectin-like domains, indicating a potential role for glycanbinding.For se v er al Cry pr oteins, inter action with sugars, gl ycopr oteins, gl ycolipids, and competition by lectins has been demonstrated in receptor binding, but a role in toxicity is not always clear.For other members of the Cry family, these studies are absent, suggesting an important gap in our knowledge that should be addressed.While for some proteins, such as Cry4B, abo ve , binding to protein receptors appears to be glycosylation independent, the potential carbohydrate-binding properties of D-II and D-III ma y pla y a r ole in binding to gl ycolipid moieties in the tar get cell membrane (as shown for Cry5B).The structural differences in glycoconjugates between insects , nematodes , and mammals is a mechanistic explanation for target range , i.e .independent of the pr otein r eceptor and may explain why the transfection of genes for suc h r eceptors does not al ways confer susceptibility to recipient cells .T his effect will be mediated by both the specificity of the carbohydrate binding domains within the Cry proteins and the natural lipid composition of the transfected cells.Understanding the exact role of glycoconjugates can be a challenge due to the difficulty in replicating the in vivo environment of the gut target tissue-especiall y with man y studies suggesting a complex coordination of binding components is r equir ed to elicit the full spectrum of toxicity.Indeed, the majority of model data comes from cell lines, which are not target-tissue specific and BBMV binding studies in which the concentrations of receptors and lipid microdomains do not necessarily accurately reflect the in vivo environment.In addition to normal development, glycan expression can be significantl y alter ed by envir onmental pr essur es, suc h as temper atur e, infection, and dietary c hanges .T his should be considered in terms of the development of Bt tolerance in target species-wher e c hanges in gl ycan binding pr ofiles may be an indication of resistance as observed with nematocidal Cry5B.Despite these experimental complexities, it is clear that glycan moieties might be critical for exerting insecticidal and nematocidal activity, with glycan-moieties observed as primary receptors critical for activity, and in more additiv e r oles that can affect the spectrum/potency of activity.Despite many years of study of the Cry proteins, our understanding of their glycoconjugate interactions r emains underinv estigated and in its infancy.Application of the tools of glycobiology to the study of insecticidal proteins in future will help us to r esolv e the importance of these interactions.
e .only proposed to occur in limited tissues and (patho)physiological conditions (Chatterjee et al. 2019 , Parker et al. 2021 )-unlike the constitutive and ubiquitous utilization of this pathway in inv ertebr ates.

Figure 1 .
Figure1.Ov ervie w of N-glycosylation in insects and nematodes.In insects and nematodes, high-mannose oligosaccharides are attached in the ER to consensus Asn residues and are subsequently processed by glycosidases and glycosyltransferases to generate a variety of N-linked structures .T he synthetic pathway shown begins with the ER gl ycosidase-pr ocessed high mannose glycan (Man 5 GlcNAc 2 ).N-link ed di v ersity is limited thr ough the expression of a hexosaminidase (Fdl in Drosophila , hex-2 C. elegans ) generating paucimannose-one of the most predominant N-linked glycans in all c har acterized insecta.A common feature of insecta N-glycans-core fucosylation, occurs at C3 and/or C6 of the reducing terminal GlcNAc (via FucT6 and FucTA in Drosophila ).N-linked diversity is expanded through the expression of less well-defined glycosyltransferases (and potentially sial yltr ansfer ases).Significant div ersity and unique gl ycan signatur es hav e been noted in complex N-gl ycans between differ ent species .For example , nematode N-glycans can contain structures (i) with a bisecting galactose, (ii) with multiple phosphocholine (PC) residues as antennal modifications, and (iii) that are fucose rich with O-methylation (Me) modifications and the extension of core fucosylated residues.Nematode structures are based on figur es fr om P asc hinger et al. (2008),Haslam et al. ( 2002 ), andWilson and P asc hinger ( 2016 ).Dipter an N-gl ycans with example antennae modifications (as found in Aedes aegypti , Anopheles gambiae , and D. melanogaster ; Kurz 2015 , P asc hinger and Wilson 2020 ) and an L. dispar zwitterionic lepidopteran N-gl ycan (P asc hinger andWilson 2020, Stanton 2017 ) show featur es fr equentl y found in man y insect species: sulphated r esidues (S ±) and glucur onic acid attached to a galactose residue.Glycans are depicted according to the Symbolic Nomenclature for Glycans, as shown in KEY , linkages are shown next to the bonds, and known enzymes are named next to initial N-glycan trimming stages.

Figure 2 .
Figure 2. Ov ervie w of O-glycan diversity in insects and nematodes .T he first residue attached to the serine/threonine determines the type of O-glycan (A) O-Xyl, (B) O-Glc, (C) O-GalNAc or mucin-type, (D) O-Man, (E) O-GlcNAc, and (F) O-Fuc.Mucin-type O-glycans (C) appear to be the most common glycans in studied insecta and nematodes with core 1 and core 1 modified with glucuronic acid (GlcA) generally most prevalent.Examples of Drosophila biosynthetic pathways illustrate some of the known O-glycan diversity with frequent sulphation and glucuronylation.(G) Examples of more complex structures and modifications including, (i) a sulphated (S) HexA-Hex-HexNAc repeat from Sf9 cells (Gaunitz 2013 ), (ii) a glycosaminoglycan-like zwitterionic glycan from the Oesophagostomum dentatum nematode modified with PC, and (iii) a Hex-HexNAc containing O-glycan modified with phosphoethanolamine ( P Etn), present in both Ae.aegypti and An.gambiae larv ae.Gl ycans ar e depicted according to the Symbolic Nomenclatur e for Glycans, as shown in KEY .

Figure 3 .
Figure 3. Gener al structur es of Sphingolipids.Sphingolipids are a class of lipids, which contain a backbone of sphingoid bases which is N-acylated with various fatty acid chains.(A) A sphingoid base composed of a 16-carbon backbone (C16, hexadecanoylsphinganine).(B) A ceramide, consisting of a sphingoid base backbone amide linked to a fatty acid.(C) A sphingomyelin, a phosphocholine headgroup attached to a ceramide.(D) Ceramide phosphoethanolamine (CPE), a phosphoethanolamine headgroup attached to a ceramide.
D) (P ane vska et al. 2019 ), whereas onl y tr ace amounts of CPE hav e been detected in mammalian cells (Bickert et al. 2015 ) and Nematoda (Satouchi et al. 1993 ) which, instead, fav our sphingomy elin synthesis (a ceramide with a phosphocholine group; Fig. 3 C).Distinct biophysical properties hav e been observ ed betw een sphingomy elin and CPE in terms of membr ane-order par ameters (Ter ov a et al. 2005 , Bjorkbom et al. 2010 ) and the ability to interact with cholesterol and form lipidrafts (Ramstedt and Slotte 2006 , Bjorkbom et al. 2010 ), suggesting they have differing biological roles (Daw alib y et al. 2016 ).It may be that these GSL backbone differences play a part in determining binding specificity of insecticidal proteins, yet, to the best of our knowledge, this has not been investigated.

Figure 5 .
Figure 5. Crystal structure of Cry1Ac in complex with GalNAc (PDB 4ARY) and a summary of notable residues identified through mutational and crystallogr a phy studies.(A) Cry1Ac shares the conserved three-domain structure of the Cry family of toxins.Domain I (orange) comprises an α-helix bundle, domain II (blue) comprises three β-sheets forming a β-prism, and domain III (green) comprises two antiparallel β-sheets forming a 'jellyroll' domain.(B) Residues (shown as sticks) implicated as significant for APN binding and/or Cry1Ac toxicity against L. dispar, M. sexta, and H. virescens in Domain III.The binding of GalNAc (yellow) r elativ e to these residues is also shown.(C) Residues that interact with GalNAc (yellow) were identified using PDBePISA and are shown as sticks (green/c y an).Residues that have also been implicated as significant for APN binding are coloured c y an.Hydrogen bonds are formed between Cry1Ac Gln509, Arg511, Arg590 and GalNAc.
. Div er gent outcomes between these experiments ar e partiall y explained by the use of different experimental systems conferring differing posttranslational modifications-especially when we know the gut tissue is the in vivo tar get.Carr oll et al .( 1997 ), first proposed a GalNAc sensitive and a GalNAc insensitive Cry1Ac binding mechanism within

Figure 6 .
Figure 6.Predicted N-and O-linked glycosylation sites of lepidopteran APNs.Re presentati ve le pidopteran APN sequences that have been reported as putative Cry toxin receptors were taken from Fonseca et al. ( 2015 ), and predictions for the number of putative N-glycosylation sites (N) and O-glycosylation sites identified using the NetNGlyc 1.0 and NetOGlyc 4.0 servers (DTU Bioinformatics), respectively (table).Species abbreviations: Ha, Helicoverpa armigera ; Hv Helicoverpa punctigera ; Se, Spodoptera exigua ; Px, Plutella xylostella ; Bm, Bombyx mori ; Ms, Manduca sexta ; Ld, Lymantria dispar ; Sl, Spodopter a litur a ; Aj, Achaea janata ; and Ep, Epiphyas postvittana .Genbank accession numbers ar e shown for eac h pr otein.To visualize the placement of N-gl ycosylation sites, m ultiple sequence alignment was pr oduced using Geneious.Mean pairwise identity is shown at the top of the alignment-gr een indicates 100% identity, y ello w indicates between 30% and 100% identity, and red indicates below 30% identity.Green arrows show the location of pr edicted N-gl ycosylation sites (lar ger arr ows ar e due to ga ps in the sequence alignment).

Table 1 .
Specificity of lectins used commonly in lectin binding studies.Ac (Cro wley et al. 1984 ) * > denotes the binding affinity where a single lectin can bind different linkages.

Table 2 .
Summary of m uta genesis studies implicating a r ole for Cry1Ac r esidues in APN binding and/or Cry1Ac toxicity against L. dispar , M. sexta , and H. virescens .